Transmission Electron Microscopy and Electron Diffraction

In TEM samples prepared by dispersing powder on Cu grids with carbon film, all plate-like crystallites exhibited approximately the same orientation of the stacking direction, perpendicular to the grid, as it is typical for layered compounds. This, of course, impedes the determination of the stacking periodicities. Such texture effects are not significant if the powder is embedded in glue, where crystallites with a translation period of ~14 Å, typical for monoclinic SrSi2O2N2, could easily be found. TEM-EDX yielded an average formula of Sr0.98Eu0.02Si2.01(8)O2.3(5)N1.6(3) for the crystallites investigated. Traces of SrSiO3 were found, although the corresponding reflections cannot be observed in PXRD data. Various SAED patterns showing translation periods of 14 Å can be simulated based on the results from single-crystal analysis. Calculated tilt angles between different zone axes correspond to the expected ones (Figure 4).[44] _{Experimental SAEDs contain 0k0 reflections with k = 2n + 1} (kinematically absent for 21 || [010]) because of dynamic effects.

Figure 4. Experimental SAED patterns (top) of monoclinic SrSi2O2N2:Eu2+ with the corresponding zone axes and simulated ones (bottom, calculated from single-crystal data). Experimental tilt angles (black) between zone axes match calculated ones (gray). The [101] pattern was recorded using another crystallite due to the limited tilt range of the sample holder.

Although the monoclinic model for SrSi2O2N2 is appropriate to simulate experimental SAED images, the local stacking sequence might differ since the beam diameter for SAED patterns (Ø = 100 nm) is significantly larger than the smallest area (10 unit cells ≈ 10 nm) that may be described by an ordered structure model. Figure 5 shows an HRTEM image of a crystallite fringe. As the Fourier transform (FT) of the marked area (diagonal 10 nm) corresponds to the SAED pattern, this area is representative for the whole area contributing to SAED patterns; no or few defects are expected.

Figure 5. (a) HRTEM image of monoclinic SrSi2O2N2:Eu2+ (zone axis [101]); approximately 10 unit cells (~10 nm) along stacking direction (white frame), the corresponding FT is shown (black frame); (b) experimental SAED pattern (beam diameter ~100 nm); (c) SAED pattern simulated using SrSi2O2N2 (monoclinic) single- crystal data.

HRTEM image simulations using the multislice method correlate the structure model to experimental images of a defocus series, passing the Scherzer defocus.[44] The defocus value in Figure 6a (zone axis [101]) is -52 nm, i.e. close to the Scherzer defocus so that contrasts can directly be correlated to atom positions.

Figure 6. HRTEM (accelerating voltage = 300 kV) images of monoclinic SrSi2O2N2:Eu2+ crystallites of zone axis [101] (a-d) and [101] (e,f) with inserted image simulations (a: Δf = -52 nm; b: Δf = -72 nm; c: Δf = -92 nm; d: Δf = -112 nm; e: Δf = +19 nm; f: Δf = +40 nm; for all simulations: aperture diameter = 20 nm-1_{, c}

s = 1.2 mm, spread of focus = 2.14 nm, beam semi-convergence = 0.60 mrad, layer thickness approx. 4 nm).

The simulation fits all features of the experimental image quite well, which is also true for the other HRTEM simulations in Figure 6. In summary, the results of TEM investigations confirm the existence of the monoclinic stacking variant of SrSi2O2N2 derived from single- crystal X-ray diffraction analysis.

2.2.3.6 Luminescence

The emission wavelength of green (presumably) triclinic SrSi2O2N2:Eu2+ (no additional reflections in PXRD pattern belonging to monoclinic modification) was reported in a range of ~537-540 nm for doping with 2 mol% Eu,[4,17,21,24,27,53,54] which leads to high quantum efficiency (QE > 90%).24 Regarding the enhancement of color rendition of white-light pc-LEDs that make use of mixtures of green and red emitting phosphors, a shift of the peak emission wavelength towards shorter wavelengths (~ 530 nm) would be desirable.[14] _The color point of “triclinic” SrSi2O2N2:Eu2+ can be changed by variation of the Eu-doping level[19,26,28-30,34] or substitution of Sr by Ca or Ba.[11,23,24,26,34,55-57] Nevertheless, all changes of the host-lattice composition shift the emission wavelength towards smaller energies. For Sr1-xBaxSi2O2N2:Eu2+ with x ≥ 0.75, a shift to higher energies can be achieved; however, its emission spectrum is located in the blue-green spectral region due to the different structure type.[28,46] The determination of luminescence properties of monoclinic SrSi2O2N2:Eu2+ cannot be done using powder samples because all obtained samples were inhomogeneous (see section: Rietveld refinement). In order to avoid averaging of emission signals of both modifications, the emission spectrum of the single crystal, which was already used for structure analysis, was measured (Figure 7).

Figure 7. Emission spectrum of the SrSi2O2N2:Eu2+ (2 mol% Eu) single crystal (λexc = 420 nm, λem = 532 nm, fwhm ~2600 cm-1_{); CIE color coordinates: x = 0.314, y = 0.621.}

The peak position was determined at λem = 532 nm by exciting with UV to blue radiation. This means that the emission wavelength of the single crystal of monoclinic SrSi2O2N2:Eu2+ is shifted at least 5 nm towards smaller wavelengths in comparison to powder material of SrSi2O2N2:Eu2+ which do not show additional reflections in PXRD pattern belonging to monoclinic modification. In order to prove whether this fact is intrinsic or caused by reabsorption of emitted high-energetic radiation in the powder sample (i.e. excitation of another Eu2+-ion by re-emitted photons due to overlap of absorption and emission band), various crystallites (independent of crystal symmetry) with anisotropic morphology were investigated. If the emission wavelength is affected by the length of the radiation pathway through a crystal (i.e. number of activator centers along the pathway, maximal for macroscopic powder sample) different values for λem are expected for varying orientations of crystallites. The more centers are involved, the more the macroscopically composed emission signal gets shifted to smaller energies. In the present case, λem was constant for various orientations of three investigated anisotropic crystallites with different sizes. Thus, the red- shifted emission wavelength of SrSi2O2N2:Eu2+ powder samples (“triclinic” modification, i.e. no additional reflections in PXRD patterns belonging to monoclinic modification), compared to SrSi2O2N2:Eu2+ single crystal (monoclinic modification), is not caused by reabsorption effects. Therefore, the observed differences in measured λem-values for SrSi2O2N2:Eu2+ (single crystal, monoclinic) and SrSi2O2N2:Eu2+ (powder, no additional reflections in PXRD pattern belonging to monoclinic modification) are significant. In order to draft a possible reason for the shift on the basis of the crystal structures, we focus on lattice parameters of triclinic and monoclinic modification. In contrast to the triclinic phase, monoclinic SrSi2O2N2 has larger a and c lattice parameters which represent the periodicity of silicate layers because corresponding settings are equal for both crystal structures. As a direct consequence, interatomic distances increase which corresponds to less corrugated chains of condensed SiON3 tetrahedra. Furthermore, Sr-O distances are also increased which should lead to a decreased 5d-orbital splitting in case of substitution of Sr by an Eu activator ion and an increase of the energetic separation of the 4f75d0 and 4f65d1 states, equivalent to a blue-shifted maximum of the emission band for monoclinic SrSi2O2N2:Eu2+ compared to the triclinic modification. For both modifications, average activator-ligand distances are slightly longer than the sum of the ionic radii.[52] This hardly leads to lattice relaxation in the case of excited Eu2+. As a consequence, less electron-phonon coupling may result in reduced Stokes shift and narrower fwhm, which decreases thermal quenching of luminescence.[58] _Luminescence properties of the above-mentioned SrSi2O2N2:Eu2+ _{powder (mixture of triclinic and}

monoclinic phase) were also measured while reabsorption effects for the powder were minimized by extrapolating the emission properties of a dilution series (silicone suspensions) to zero phosphor concentration, in order to ensure comparability of single-crystal and powder data. An emission wavelength of λem = 535 nm was determined supporting the thesis that the emission band of the monoclinic modification is blue-shifted in comparison to samples without additional reflections in the PXRD pattern (537-540 nm).