Skip to content

Chemometric Raman Spectral Analysis of Diamond-like Carbon Films Deposited by Plasma Discharge

MetadataDetails
Publication Date2024-06-25
JournalJournal of Photopolymer Science and Technology
AuthorsYoshihisa Osano, Hiroyuki Fukue, Susumu Takabayashi, Shinsuke Kunitsugu, Yuichi Imai
InstitutionsOkayama University of Science, Engineering Systems (United States)
Citations2
AnalysisFull AI Review Included

Chemometric Raman Spectral Analysis of Diamond-like Carbon Films

Section titled “Chemometric Raman Spectral Analysis of Diamond-like Carbon Films”

This study introduces a chemometric approach using Nonlinear Least Squares (NLS) fitting combined with a pseudo-Voigt function to automate and standardize the five-peak separation analysis of Diamond-like Carbon (DLC) Raman spectra.

  • Core Problem Solved: Conventional two-peak fitting is structurally limited, and manual five-peak fitting is slow, requires high expertise, and lacks reproducibility.
  • Methodology: Automated five-peak separation analysis using the Levenberg-Marquardt NLS algorithm and a modified pseudo-Voigt function.
  • Computational Efficiency: The pseudo-Voigt function successfully replaced the computationally complex Voigt function (R2 = 0.99996), reducing execution time by over 80% (0.110 s vs. 0.575 s).
  • Accuracy Improvement: Automated NLS fitting consistently achieved higher fitting accuracy (higher R2 and lower chi-squared, X2) compared to conventional manual fitting for both nonhydrogenated (HF-HiPIMS) and hydrogenated (AC-HV-CVD) DLC films.
  • Engineering Impact: The automated process performs over 100 calculations in less than 3 seconds, providing highly precise, reproducible, and fast structural analysis results independent of the analyst’s skill level.
  • Structural Insight: The method facilitates consistent interpretation of the five active Raman bands (N, D, G-, G+, D’), which correlate strongly with the sp3 C-C/(sp3 C-C + sp2 C=C) ratio.
ParameterValueUnitContext
Deposition Method 1HF-HiPIMSN/ANonhydrogenated DLC
Target MaterialGraphite (3-inch diameter)N/AHF-HiPIMS
Operating Pressure0.5PaHF-HiPIMS
Applied Voltage (Negative)-810VHF-HiPIMS
Waveform Frequency200HzHF-HiPIMS
Deposition Method 2AC-HV-CVDN/AHydrogenated DLC
AC Voltage5kVAC-HV-CVD
Offset Voltage2kVAC-HV-CVD
CH4 Gas Flow Rate96sccmAC-HV-CVD
Operating Pressure39PaAC-HV-CVD
Raman AnalysisN/AN/ASpectrometer (Raman-11)
Laser Wavelength532nmExcitation source
Output Power0.5mWLaser power
Spot Diameter2.55”mMeasurement area
Fitting Accuracy (HF-HiPIMS)
Conventional Manual R20.98874N/AVoigt function
Automated NLS R20.99486N/APseudo-Voigt function
Fitting Accuracy (AC-HV-CVD)
Conventional Manual R20.97583N/AVoigt function
Automated NLS R20.97624N/APseudo-Voigt function
Execution Time Comparison
Voigt Function (Avg)0.575sConventional method
Pseudo-Voigt Function (Avg)0.110sNew automated method

The study utilized two distinct plasma deposition methods to create DLC films and applied a chemometric approach for spectral analysis.

  • Nonhydrogenated DLC (HF-HiPIMS):
    • Used a 3-inch graphite target with Ar sputtering gas (5 sccm).
    • Substrate rotation: 5 rpm.
    • High-frequency inclusion high-power impulse magnetron sputtering (HF-HiPIMS) applied a negative voltage of -810 V at 200 Hz.
    • Deposition occurred at 0.5 Pa operating pressure for 2 hours.
  • Hydrogenated DLC (AC-HV-CVD):
    • Used alternating current high voltage burst plasma chemical vapor deposition (AC-HV-CVD).
    • Process gas: CH4 (96 sccm).
    • Electrical parameters: 5 kV AC voltage, 2 kV offset voltage, 10 kHz frequency.
    • Deposition occurred at 39 Pa operating pressure for 1 hour.

The analysis followed a standardized, automated procedure using proprietary Python code:

  1. Data Preparation: Load spectral data, define the analysis range, and apply the differential spectrum method to identify initial peak positions.
  2. Preprocessing: Perform baseline removal and normalization of the spectra.
  3. Fitting Algorithm: Execute Nonlinear Least Squares (NLS) fitting using the Levenberg-Marquardt algorithm.
  4. Function Selection (Pseudo-Voigt): The complex Voigt function (Vj), which involves a convolution integral, was replaced by a modified pseudo-Voigt function (Pj).
    • Pj is a linear combination of Gaussian (Gj) and Lorentzian (Lj) functions.
    • The Breit-Wigner-Fano (BWF) function (B3) was applied exclusively to the G band (j=3) to account for asymmetry, though it simplifies to Lorentzian for insulating DLC films (where the asymmetry parameter q is infinite).
  5. Five-Peak Separation: The NLS method automatically optimized parameters for the five active Raman bands (N, D, G-, G+, D’) derived from structural symmetry operations.
  6. Validation: Fitting accuracy was confirmed by checking the coefficient of determination (R2) and chi-squared (X2) values, demonstrating superior performance compared to conventional manual fitting.

The ability to rapidly and accurately characterize the sp3/sp2 ratio and structural properties of DLC films is critical for quality control and material development in several high-performance industries:

  • Automotive Industry: DLC films are used for engine components and moving parts due to their low coefficient of friction and high hardness, improving fuel efficiency and lifespan.
  • Medical Devices: DLC’s biocompatibility makes it suitable for coatings on implants, surgical tools, and other medical devices.
  • Packaging Industry: DLC films provide excellent gas barrier properties, particularly for polyethylene terephthalate (PET) bottles, extending the shelf life of beverages.
  • Industrial Coatings: Applications requiring high wear resistance, such as tools, molds, and precision components.
  • Materials Research and Development (R&D): The automated, reproducible analysis method accelerates the development cycle for new DLC formulations by providing consistent, high-precision structural feedback.
View Original Abstract

In this study, a chemometric approach for five-peak separation analysis of the Raman spectra of diamond-like carbon (DLC) films was investigated. DLC films were deposited by high-frequency inclusion high-power impulse magnetron sputtering and alternating current high voltage burst plasma chemical vapor deposition. We used the pseudo-Voigt function as an alternative to the conventional Voigt function and applied the nonlinear least squares method. The results not only facilitate automated analysis but also guarantee highly accurate results regardless of the analyst’s level of expertise. This approach is expected to lead to consistent interpretation of Raman spectral analysis of DLC films and further research and understanding of the properties of DLC films.