Raman Microspectroscopy for Real-Time Structure Indicator in Ultrafast Laser Writing

TL;DR

This work addresses the lack of real-time feedback in fs-laser diamond fabrication by employing Raman microspectroscopy to monitor graphitization. It demonstrates that depletion of the sp3 diamond line at $1332~\mathrm{cm^{-1}}$ correlates monotonically with electrical resistance in laser-written graphitic electrodes, while sp2 intensity alone is less reliable at high dose. The authors introduce multi-band metrics, such as the sp2/sp3 ratio, and hyperspectral unmixing to quantify phase fractions without predefined bands, providing robust indicators of process quality. The approach is shown to be transferable to other host materials, offering a practical route toward specification-driven fs-laser microfabrication with in-situ monitoring capabilities.

Abstract

Femtosecond laser fabrication enables the creation of a wide range of devices, but its scalability and yield can be limited by the lack of real-time, in-situ monitoring tools. In particular, there is a strong need for metrics that directly correlate with device performance. Raman microspectroscopy provides a non-destructive route for in-situ characterization. Here, we demonstrate its potential to assess the electrical performance of laser-written graphitic electrodes in diamond. By combining hyperspectral mapping with electrical testing, we show that depletion of the 1332 cm$^{-1}$ sp3 Raman line serves as a monotonic and robust predictor of resistance, offering clear advantages over commonly used spectral features. We further introduce hyperspectral unmixing as a label-free approach to identify relevant spectral signatures in fabrication processes where Raman markers are less defined. Importantly, the methodology we present is not restricted to diamond but can be adapted to other host materials and functionalities, offering a practical path toward specification-driven fs-laser microfabrication.

Figures (17)

• Figure 1: Laser-written graphitic structures with optical/PL/Raman characterization.a) Schematic of fs-laser fabrication of graphitic pads and wires. The fabrication beam shares a common optical path with the 532 nm Raman excitation, and the back-scattered Raman signal is collected to enable in situ monitoring and post-fabrication mapping. b-c) Optical micrographs, PL maps, and Raman intensity maps (sp3 and sp2 windows) for wires written at scan speeds of 20 $\mu$m s$^{-1}$ and 200 $\mu$m s$^{-1}$, respectively, collected after fabrication. For direct comparison, PL and Raman maps are unprocessed and share identical count limits. The 20 $\mu$m s$^{-1}$ wire exhibits a resistance of $46.2~\mathrm{k}\Omega \pm 41.2~\Omega$, whereas the 200 $\mu$m s$^{-1}$ wire exhibits $2.68~\mathrm{M}\Omega \pm 836~\Omega$. The stronger optical darkening at the lower scan speed provides a first qualitative indication of increased graphitization. However, the contrast in optical micrographs and PL maps alone is difficult to resolve quantitatively. By contrast, Raman mapping of the sp3 window provides a direct and more reliable measure of the degree of graphitization, offering phase-specific insight that complements the broadband PL response and sp2 contrast.
• Figure 1: Laser-written graphitic structures with optical/PL/Raman characterisation.a-c) Optical micrographs, PL maps, and Raman intensity maps (sp3 and sp2 windows) for pads written at scan speeds of 100 and 200 $\mu$m s$^{-1}$, respectively. The 100 $\mu$m s$^{-1}$ pad exhibits a resistance of $214~\Omega \pm 10.4~\Omega$, whereas the 200 $\mu$m s$^{-1}$ pad exhibits $3.69~\mathrm{k}\Omega \pm 1.59~\Omega$, i.e., a difference exceeding one order of magnitude. The optical darkening and PL contrast are effectively indistinguishable between the two speeds; thus, optical micrographs and PL maps cannot resolve electrodes that differ in resistance by more than an order of magnitude. By contrast, Raman mapping of the sp3 window provides a direct and more reliable measure of the degree of graphitization, offering phase-specific insight that complements the broadband PL response and the lower-contrast sp2 map.
• Figure 2: Spectroscopic analysis of laser-written graphitic structures at different scan speeds.a-b) Representative normalized Raman spectra (532 nm excitation) from pristine diamond and a fully graphitized region (see \ref{['sec:Method']} and Supporting Note 2 for normalization and pixel selection). The pristine diamond spectrum is dominated by the sp3 line at 1332 cm$^{-1}$. In the graphitized region the sp3 signal is strongly suppressed, while a broad disorder-activated D band appears near 1350 cm$^{-1}$ together with the sp2 G peak centered around 1580 cm$^{-1}$, indicative of local graphitization. c) Scan-speed-dependent mean Raman spectra restricted to the sp3 (diamond) and sp2 (graphitic) windows, shown for wires written at 10–1270 $\mu$m s$^{-1}$ (top to bottom). For each speed, the mean is computed from 60 spectra per wire extracted from the hyperspectral maps, revealing progressive depletion of sp3 features and growth of the sp2 response with decreasing laser scan speed. d) Speed-dependent quantitative summary: integrated intensities (from 60 spectra per wire extracted from the hyperspectral maps) of the sp3 and sp2 windows, together with the sp2/sp3 ratio, plotted against scan speed, error bar indicating one sigma error.
• Figure 2: CW-laser-assisted removal of graphitic debris from laser-written pads. Optical micrograph of a graphitic pad after femtosecond (fs) laser writing. The dashed red box highlights debris generated during fs writing; the dashed green box marks a region subsequently scanned with a CW laser, where the surface graphitic debris has been effectively removed, revealing the underlying written structure.
• Figure 3: Evaluation of spectral unmixing for electrode surface graphitization.a-b) PL intensity map and normalized first-order Raman intensity (sp3 line at 1332 cm$^{-1}$) for a graphitic wire written at 50 $\mu$m s$^{-1}$ (74.8 k$\Omega$). c) Map of sp3-to-sp2 ratio. d) Threshold-based segmentation by the sp3-to-sp2 ratio $\rho = I_{1332}/I_{1580}$. (d.1) Segmentation map using experimentally determined thresholds for non-graphitized ($\rho\leq3$), partially graphitized ($3<\rho<9$), and fully-graphitized ($\rho\geq9$). (d.2) Pixel distribution of $\rho_{sp3}$ over the complete set of graphitic wires spectral data (i.e., including all different writing speeds), depicting selected thresholds. (d.3) Class-averaged spectra (mean $\pm$ 1 s.d.) for the 74.8 k$\Omega$ electrode, showing the sp3 line (1332 cm$^{-1}$), D band ($\sim$1350 cm$^{-1}$), NV$^{0}$ fluorescence ($\sim$1420 cm$^{-1}$), and G band ($\sim$1580 cm$^{-1}$). e) Spectral unmixing by Vector Component Analysis (VCA) with Fully Constrained Least Squares (FCLS) over spectral region of interest (ROI) 1 (1000-1800cm-1): FCLS abundance maps for endmember 1 (e.1) and endmember 2 (e.2), and corresponding endmember spectra with band markers (e.3). f) Same unmixing workflow over ROI 2 (1345–1800 cm$^{-1}$, excluding the 1332 cm$^{-1}$ line): abundance maps for endmember 1 (f.1) and endmember 2 (f.2), and endmember spectra (f.3).