# Fileset

[quasi_bulk_ScN_opt_props_Suppl_v3Revised.pdf](https://mdr.nims.go.jp/filesets/7cd236d1-bfed-449b-a2fc-fab32773da14/download)

## Creator

Jona Grümbel, Rüdiger Goldhahn, Martin Feneberg, [Yuichi Oshima](https://orcid.org/0000-0001-8293-4891), Adam Dubroka, Manfred Ramsteiner

## Rights

©2024 American Physical Society[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Band gaps and phonons of quasi-bulk rocksalt ScN](https://mdr.nims.go.jp/datasets/cd92f0a9-f7dd-4e1b-bbf1-9c83ca6d60b6)

## Fulltext

Supplement: Band gaps and phonons of quasi-bulk rocksalt ScNJona Grümbel,∗ Rüdiger Goldhahn, and Martin FenebergInstitut für Physik, Otto-von-Guericke-Universität Magdeburg, Universitätsplatz 2, 39106, Magdeburg, GermanyYuichi OshimaResearch Center for Electronic and Optical Materials,National Institute for Materials Science , 1-1 Namiki, Tsukuba, Ibaraki 305-0044, JapanAdam DubrokaDepartment of Condensed Matter Physics, Masaryk University, Kotlá°ská 2, 611-37 Brno, Czech RepublicManfred RamsteinerPaul-Drude-Institut für Festkörperelektronik (PDI), Hausvogteiplatz 5-7, 10117, Berlin, GermanySI. EXPERIMENTAL SETUPSFor our ellipsometry/re�ectivity measurements we use three di�erent machines, where the WVASE32 ellipsometry software(Woollam) was used for both data recording and analyzing:(I) A Wollam VASE rotating compensator UV ellipsometer for 0.5−6.5 eV with a HS190 monochromator and a highpressure Xe-lamp as light source. The spectral resolution trace is shown in Fig. S3.(II) A Woollam VASE fourier transform IR ellipsometer for 280−6000 cm−1 (35−750meV) with a Carbon Globar drivenMichelson-Morley Interferometer as light source. The spectral resolution (energy step width) is set to 2 cm−1.(III) A Bruker Vertex 70V spectrometer for 60−700 cm−1, equipped with a Hg lamp as source and DTGS detector. Thespectral resolution is set to 1 cm−1.Raman measurements are recorded with two di�erent machines:(I) A Tri-Vista Raman microscope with an incident laser wavelength of 532.1 nm, an Olympus BX50 microscope with ax50 objective for focussing, a Princeton Instruments SP2750i monochromator with a focal length of 750mm, gratingparameter of 1800mm−1, slit width of 200 µm and a peltier-cooled charge coupled device (CCD) camera (2048pxhorizontal).(II) A Horiba LabRam HR evolution system with incident laser wavelength of either 632.8 nm, or 472.9 nm, a x50 objectivefor focussing, a focal length of 800mm, grating parameters of 1800mm−1 or 600mm−1, slit width of 200 µm and alN2-cooled Symphony CCD camera (1024px horizontal).Di�erent setups used to record Raman/PL spectra result in the following spectral resolutions ∆krel at 300 cm−1 (∆λ ≈slit width/(grating× focal length)):Setup 1: 633 nm, grating 1800mm−1: ∆krel = 3.3 cm−1Setup 2: 532 nm, grating 1800mm−1: ∆krel = 4.8 cm−1Setup 3: 473 nm, grating 1800mm−1: ∆krel = 6.0 cm−1Setup 4: 473 nm, grating 600mm−1: ∆krel = 18 cm−1 (2meV)Setups 1-3 are used only for Raman measurements (Fig. 4a of the main article) and Setup 4 for the combined Raman/PLmeasurement (Fig. 4b of the main article).∗ jona.gruembel@ovgu.demailto:jona.gruembel@ovgu.de2(a) (b)Figure S1: Parametric semiconductor (PSEMI) oscillator model �ts applied to (a) Ψ and (b) ∆ of UVSE beforepoint-by-point �tting.SII. ELLIPSOMETRY PROCEDUREIn the NIR/VIS/UV spectral range we model two layers: (I) a surface roughness layer, which we model by using aBruggemann e�ective-medium layer with a void/layer ratio of 50% and a variable layer thickness and (II) the ScN �lmlayer, which is constructed by various oscillators as shown in Fig. 3b in the main article. Due to the large thickness ofour ScN �lms we treat them as bulk material and therefore apply no substrate model. In Figs. S1a and S1b it is obvious,that already the applied model matches the experimental data nearly perfectly and hence, since the WVASE model func-tions are inherently Kramers-Kronig-consistent, the NIR-UV dielectric function determined from point-by-point �tting isKramers-Kronig-consistent. In the IR/FIR spectral range only a single layer for the ScN �lm is used. The bulk approachis further supported by the absence of Fabry-Perot-oscillations in the NIR/VIS range and the absence of sapphire substratecontributions in the MIR range (see Figs. 2a and 2b in the main article).SIII. RAMAN SPECTRA LINE SHAPE FITSFor quantitative analysis we determine the energy positions by line shape �ts. There, we deal with two problems: (I)the underlying luminescence signal and (II) the non-Lorentzian line shape of some signals. The non-Lorentzian line shapecan either arise from coupling with free charge carriers or from even number TO phonons. As discussed in the mainarticle, previous XRD results suggest that those signal possibly stem from the even number TO lines because they exhibit2ωTO ≈ ωLO. The �rst order Raman scattering intensity typically exhibits Lorentzian type line shapes, given byI(ω) =Aγ24(ω − ω0)2 + γ2(1)with amplitude A, eigenfrequency ω0, and broadening γ. We apply a local linear background, �tting each peak individuallyfor best �t results. In total, we haveI(ω) = I0 +mω +Aγ24(ω − ω0)2 + γ2(2)where m is an arbitrary but linear slope and I0 a constant background.In Figs. S2a-S2h �t results of all evaluated phonon lines are shown. The applied model matches well with the data, onlyfor the 2LO signal small deviations are visible.SIV. ERRORSFor values directly extracted as �t parameters the given errors in Tabs. 1 and 2 of the main article display the 99.5%con�dence bounds of the �t. For the derived parameters ε∞, εstat, Z∗, ω(LST)LOand the LST-relation probe, the errors are3(a) (b)(c)(d) (e) (f)(g) (h)Figure S2: Fit results at nLO/TO-spectral regions of the Raman spectrum with ELaser = 2.62 eV.calculated using linear error propagation law and the corresponding �t errors of the required parameters. For the transitionenergies EΓ and EΓ′ , which display via spline�t determined in�ection points of ε2, the error is set as the spectral resolutionof the UV ellipsometer at the corresponding photon energy (see Fig. S3). For the indirect bandgap the error is an estimatefor our scale-reading precision when deriving it from ⟨ε1⟩.Figure S3: Spectral resolution trace of the UVSE machine. Note, that we use auto slit mode but calculated the error witha constant dslit = 500µm, which is the maximum allowed slit width.4 Supplement: Band gaps and phonons of quasi-bulk rocksalt ScN SI. experimental setups SII. ellipsometry procedure SIII. Raman spectra line shape fits SIV. errors References