sergey kucheryavski [email protected] raman spectroscopy acquisition, preprocessing and analysis of...
TRANSCRIPT
Sergey Kucheryavski
Raman spectroscopy
Acquisition, preprocessing and analysis of spectra
Raman spectrometer scheme
Acquisition, preprocessing and analysis of Raman spectra 2
Credits: http://www.doitpoms.ac.uk/tlplib/raman/method.php
Raman spectrometer
Acquisition, preprocessing and analysis of Raman spectra 3
Raman spectrometer
Acquisition, preprocessing and analysis of Raman spectra 4
Probes and fibers
Acquisition, preprocessing and analysis of Raman spectra 5
Non contact probe
Acquisition, preprocessing and analysis of Raman spectra 6
Acquisition of Raman spectra
Acquisition, preprocessing and analysis of Raman spectra 7
Raman signal is weak
Only around 1 in every 30 million photons is Raman scattered
Acquisition of Raman spectra
Acquisition, preprocessing and analysis of Raman spectra 9
Issues
• Cosmic rays
• Noise
• Detection limits
• Fluorescence
Parameters
• Laser frequency
• Laser power
• Exposure time
• Number of scans
Preprocessing
• Spectral truncation
• Noise reduction
• Baseline correction
• Derivatives
Preprocessing
Preprocessing – a way to improve signal for further analysis
What can be improved
– Noise reduction
– Correction of baseline
– Resolving merged meaks
– Removing physical effects
How it works:
– X’ = F(X)
– xij = fj(xij)
106. Data preprocessing
Cosmic spikes
Noise and detection limits
Fluorescence and background correction
Cosmic spikes
• occasionally appears in spectra as very narrow peaks
• caused by high energy cosmic rays
• typical issue for CCD based instruments
• most of the acquisition software include algorithms to remove the effect
Acquisition, preprocessing and analysis of Raman spectra 12
Credits: Confocal Raman Microscopy. ed. Thomas Dieing, et al.
Cosmic spikes
Noise and detection limits
Fluorescence and background correction
Noise and detection limits
CCD detectors have photon noise, dark noise and read noise
Raman signal is weak
To get a good signal/noise ratio
• cool CCD
• higher concentration
• longer exposure time
• more scans for the same sample
• de-noising preprocessing
Acquisition, preprocessing and analysis of Raman spectra 14
Noise and detection limits
CCD detectors have photon noise, dark noise and read noise
Raman signal is weak
To get a good signal/noise ratio
• cool CCD
• higher concentration
• longer exposure time
• more scans for the same sample
• de-noising preprocessing
Acquisition, preprocessing and analysis of Raman spectra 15
Acquisition parameters and concentration
Acquisition, preprocessing and analysis of Raman spectra 16
25% ethanol
t = 5s
t = 3s
t = 1s
Acquisition parameters and concentration
Acquisition, preprocessing and analysis of Raman spectra 17
10% ethanol
t = 3s
t = 1s
t = 1s
Acquisition parameters and concentration
Acquisition, preprocessing and analysis of Raman spectra 18
10% ethanol
t = 3s
t = 1s
Acquisition parameters and concentration
Acquisition, preprocessing and analysis of Raman spectra 19
1% ethanol
t = 5s
t = 3s
t = 1s
Trancating spectra
Acquisition, preprocessing and analysis of Raman spectra 20
1% ethanol
t = 5s
t = 3s
t = 1s
Acquisition parameters and concentration
Acquisition, preprocessing and analysis of Raman spectra 21
1% ethanol
t = 5s
t = 3s
t = 1s
Acquisition parameters and concentration
Acquisition, preprocessing and analysis of Raman spectra 22
Butter
t = 1s, 5 scans
t = 1s, 3 scans
t = 1s, 1 scan
Acquisition parameters and concentration
Acquisition, preprocessing and analysis of Raman spectra 23
Butter
t = 1s, 5 scans
t = 1s, 3 scans
t = 1s, 1 scan
Playing with acquisition parameters
Acquisition, preprocessing and analysis of Raman spectra 24
Butter
t = 3s, 5 scans
t = 1s, 1 scan
Using filters for noise removal
• Linear filters: moving average, gaussian
• Wavelet decomposition
• Savitzky-Golay smoothing
Acquisition, preprocessing and analysis of Raman spectra 25
w = 5 d = 1
Using filters for noise removal
Acquisition, preprocessing and analysis of Raman spectra 26
SG filtered
noised
original
Cosmic spikes
Noise and detection limits
Fluorescence and background correction
Fluorescence
Mechanism
• appears if molecules can absorb the laser
radiation at particular wavelength
• the absorbed light excites electrons to higher
energy levels
• electrons return to the ground state by emitting
light of longer wavelength
Acquisition, preprocessing and analysis of Raman spectra 28
Fluorescence
How decrease/get rid of fluorescence:
• remove impurities from solid samples
• using microprobes or confocal Raman microscopy (for solid samples)
• using lasers with wavelength in NIR range
• proper preprocessing (baseline correction)
Acquisition, preprocessing and analysis of Raman spectra 29
Features
• very common for colored (especially dark) samples
• several orders of magnitude stronger than Raman scattering
• has a broad emission
Fluorescence
Color of samples
Acquisition, preprocessing and analysis of Raman spectra 30
Laser wavelength
• Visible — higher energy, stronger signal, deeper penetration, better resolution,
fluorescence (good for inorganic materials)
• NIR — lower energy, weaker signal, worse resolution, smaller fluorescence
effect (suitable for organic materials)
Acquisition, preprocessing and analysis of Raman spectra 31
Credits: http://www.horiba,com
Baseline correction
Acquisition, preprocessing and analysis of Raman spectra 32
Baseline shift and curvature caused by noise, fluorescence, CCD
background, interference, etc.
Automatic baseline correction
Baseline correction
Acquisition, preprocessing and analysis of Raman spectra 33
Automatic baseline correction
d = 4
Baseline correction
Acquisition, preprocessing and analysis of Raman spectra 34
Automatic baseline correction
d = 6
Baseline correction
Acquisition, preprocessing and analysis of Raman spectra 35
Semi-automatic baseline correction
Conclusions
Issues
• Cosmic rays
• Noise
• Detection limits
• Fluorescence
Acquisition, preprocessing and analysis of Raman spectra 36
Parameters
• Laser frequency
• Laser power
• Exposure time
• Number of scans
Preprocessing
• Spectral truncation
• Noise reduction
• Baseline correction
• Derivatives