rt-TDDFT modeling of thermal emission by laser-heated glasses

Grigory Kolesov · 2024

Read rt-TDDFT modeling of thermal emission by laser-heated glasses by Grigory Kolesov (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-10561.txt.

Introduction

Grigory Kolesov models how hot glass glows in a 2024 paper on laser processing. When a laser heats glass, only a thin layer gets hot, from about 50 to 1000 micrometers deep.

Because the hot layer is so thin, the light it gives off need not match the glow of a fully hot body. The material's own vibration patterns can shape the spectrum. Kolesov uses a time-based computer method that follows the electrons in the glass as conditions change. From it he computes how strongly the glass emits at each wavelength. Such numbers let engineers judge how far thermal camera readings can be trusted when they check glass during laser work.

Catalog record

CREATORGrigory Kolesov
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE23 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-10561

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

rt-TDDFT modeling of thermal emission by laser-heated glasses Grigory Kolesov Abstract In the laser processing of glass, a 5̃0-1000 μ \mu m-thick layer of glass is heated to a high temperature by the laser beam. Due to the shallow depth of this hot layer, the infrared emission and absorption spectra may deviate from the black-body spectra and can be influenced by the vibrational structure of the material. Real-time time-dependent density functional theory (rt-TDDFT) modeling of the thermal radiation by such hot layers allows us to calculate the emissivity and thus to evaluate the reliability of the measurements conducted with thermal cameras at specific wavelengths. Introduction Accurate measurements of temperature of glasses close to the softening point ( T\sim 1000-2000 K) require the knowledge of emissivity at the wavelengths used by the thermal camera, which are typically in the 4-8 \mu m (mid-wavelength infrared, MWIR) and 8-14 \mu m (long-wavelength infrared, LWIR) ranges. Because in the laser processing of glasses, the laser-heated region can be arbitrarily thin, the emissivity, especially in the MWIR range, can be diminished and calibration procedures performed on the bulk glass can be invalidated. Density functional theory (DFT) simulations can provide accurate information about the infrared response, absorption, and emission of the material. To account for anharmonism and finite temperatures, the molecular dynamics (MD) Green-Kubo approach was developed bader_berne ; iftimie ; thomas2013 . One difficulty with this method is that the dipole moment is ill-defined in the periodic boundary conditions, which are most suited for simulations of glass. This difficulty was overcome through the development of the modern theory of polarization resta1992 ; king1993theory . With its use, the infrared response was previously computed with the Green-Kubo approach in Refs. iftimie ; thomas2013 via calculation of maximally-localized Wannier orbitals. Below, we demonstrate another approach which, within real-time time-dependent DFT (rt-TDDFT), directly computes the change of the macroscopic polarization as the total current through the simulation cell. We also test simplified approaches based on charge partitioning schemes. I Materials and Methods In this work, our objective is to calculate the emission and absorption of IR radiation by glass at high temperature close to the softening point. Additionally, the OH (hydroxyl) emission and absorption band is of particular interest. In order to make a comparison with experimental data it useful to calculate the spectral radiance L : L(\omega,T)=\frac{\hbar\omega^{3}}{4\pi^{3}c^{2}}\frac{1-{\text{e}}^{-\alpha(\omega)d}}{{\text{e}}^{\frac{\hbar\omega}{k_{\text{B}}T}}-1}, (1) where \alpha is the Napierian absorption coefficient and d is the thickness of the emitting layer (see Appendix A for relations between different absorption coefficients). In this work, a hot layer of uniform temperature is considered for simplicity. The IR light is absorbed in the material through vibrational excitation of atomic dipoles. The quantum dipole-dipole autocorrelation function and its Fourier transform can be approximately computed via the classical autocorrelation function bader_berne : \displaystyle\int_{-\infty}^{\infty}\differential t{\text{e}}^{-i\omega t}\langle\frac{1}{2}\left[\hat{\bm{\mu}}(0),\hat{\bm{\mu}}(t)\right]_{+}\rangle \displaystyle=\frac{\beta\hbar\omega}{2}\coth(\frac{\beta\hbar\omega}{2})\int_{-\infty}^{\infty}\differential t\ {\text{e}}^{-i\omega t}\langle\bm{\mu}(0)\cdot\bm{\mu}(t)\rangle_{\text{cl}}, (2) where []_{+} denotes the anticommutator, \bm{\mu} is the total dipole moment of the system, \hat{\bm{\mu}} is the dipole moment operator and \beta=1/k_{\text{B}}T . The imaginary part \varepsilon_{2}(\omega) of the dielectric function can be found via Green-Kubo relation for the dipole or dipole time-derivative autocorrelation function iftimie : \varepsilon_{2}(\omega)=\frac{\beta}{6V\omega\varepsilon_{0}}\int_{-\infty}^{\infty}\differential t\ {\text{e}}^{-i\omega t}\langle\dot{\bm{\mu}}(0)\cdot\dot{\bm{\mu}}(t)\rangle_{\text{cl}}, (3) where V is the system’s volume and the dot indicates the time-derivative. The time-derivative form of the relation is chosen for numerical efficiency. The real part of the dielectric function \varepsilon_{1}(\omega) is obtained from the Kramers-Krönig relation: \varepsilon_{1}(\omega)-1=\frac{2}{\pi}{\mathcal{P}}\int_{0}^{\infty}\frac{\omega^{\prime}\varepsilon_{2}(\omega^{\prime})}{\omega^{\prime 2}-\omega^{2}}\differential\omega^{\prime}, (4) with {\mathcal{P}} indicating Cauchy principal value. Eq. ( 4 ) can either be evaluated via direct numerical integration, or by taking two successive Fourier transforms frohlich ; iftimie : \varepsilon_{1}(\omega)-1=\frac{2}{\pi}\int_{0}^{\infty}\int_{0}^{\infty}\varepsilon_{2}(\omega^{\prime})\sin(\omega' \tau)\cos(\omega\tau)\differential\omega^{\prime}\differential\tau (5) In the periodic boundary conditions used in the calculations below the dipole moment is not defined. One way to deal with this problem is to use localized Wannier functions resta_review ; iftimie ; thomas2013 . However, this requires finding localized Wannier functions at each MD time step. Another approach, which we propose here, is to directly calculate the autocorrelation function for the time derivative of the macroscopic polarization \mathbf{P}(t) , which is the total current through the simulation cell resta1992 ; king1993theory ; resta_review : \mathbf{P}(t)=\frac{1}{V_{\text{cell}}}\int_{0}^{t}\mathbf{j}_{\text{cell}}(t^{\prime})dt^{\prime}. (6) \mathbf{j}_{\text{cell}} can be calculated either directly in real-time time-dependent DFT (rt-TDDFT, Method 1) or by calculation of the partial charges on atoms and multiplying them by the atomic velocity vectors (Method 2). Numerical simulations. We used the SIESTA siesta02 DFT code with certain modifications implemented by us, including rt-TDDFT kolesov2015real . We used the PBE exchange-correlation functional pbe and the double- \zeta +polarization (DZP) basis set. We found that this higher-accuracy basis is necessary to correctly reproduce the frequency of the OH stretch vibration. A fused-silica simulation cell was prepared by simulated annealing at constant volume for 20 ps. The resulting simulation cell contained 95 atoms including two hydroxyl groups. We used a time step of 0.3 fs in all our standard (adiabatic) molecular dynamics simulations. Such a reduced time step is necessary to capture motion of the hydrogen atom. In the rt-TDDFT calculations we used a time step of 1 a.u. ( \approx 24 as). In order to obtain the realistic concentration of OH we also calculated the absorption spectrum for the fused-silica cell with no OHs and then calculated the weighted average of the two spectra with weights made to match the experimental OH concentration. We used the Hirshfeld and Bader charge partitioning methods for the partial charge-based method (Method 2). We also tested the Voronoi deformation density charge partitioning scheme voronoi , however the results of this scheme were very similar to those obtained with the Hirshfeld method, so only the latter is reported here. Bader charges were calculated with Bader code badercode . II Results and discussion II.1 Fused silica. Figure 1: Method 1: The calculated emission spectra (a) and the spectral emissivity (b), for a 6mm-thick, 0.15 weight-% OH fused silica sample at T=1600 K. In (a) the orange dotted line shows the black-body emission spectrum, the thin blue line shows the calculated emission spectrum while the purple line shows the smoothed version of the calculated spectrum. The latter was used to calculate (b). In (b) the solid blue line shows the calculated spectral emissivity and the dashed orange line shows the experimental curve from Ref. dvurechensky . In both plots the red dashed lines mark the wavelengths range between 4 \mu m and 8 \mu m often used by thermal cameras. Figure 2: Method 2: The calculated emission spectra (a) and the spectral emissivity (b), for a 6mm-thick, 0.15 weight-% OH fused silica sample at T=1600 K. In (a) the orange dotted line shows the black-body emission spectrum, the thin blue line shows the calculated emission spectrum (Hirshfeld charge method) while the purple line shows the smoothed version of the calculated spectrum. The thick gray line shows the spectrum calculated with Bader charges. In (b) the solid blue line (Hirshfeld charge method, smoothed) and thick gray line (Bader charge method) show the calculated spectral emissivity. The dashed orange line shows the experimental curve from Ref. dvurechensky . Ref williams Ref dvurechensky rt-TDDFT (this work, Method 1) DFT (this work, Method 2) \alpha , cm -1 ( 10^{3} weight–ppm OH) 12.83 9.87 14.45 13.66 \epsilon_{10} , L cm -1 /mol 43 33.1 48.44

Read the complete text