Instrumentation for Raman Spectroscopy R M Erasmus Raman and Luminescence Laboratory and Materials Physics Research Institute, School of Physics, University of the Witwatersrand, Johannesburg, South Africa and DST/NRF Centre of Excellence in Strong Materials, University of the Witwatersrand, Johannesburg, South Africa
Aim of presentation
Overview of instrumentation for Raman spectroscopy
Specific focus on dispersive instrumentation. Fourier-Transform Raman spectroscopy (FT-Raman) in later presentation. Description of constituent components and their respective roles.
Performance factors to consider for Raman instrumentation
Raman photon collection efficiency
Stray light rejection
Spectral resolution
Calibration stability
Data acquisition speed
Operational flexibility
Cost
Basic layout of dispersive Raman spectrometer Polarisation control Light source
Pre-sample filter
Sample Collection optics Post-sample filter
Red components = essential
Polarisation analyser Wavelength selector (monochromator or spectrograph)
Detector
Basic layout of dispersive Raman spectrometer
Laser
Detection Device Rayleigh Filter (Notch/Edge filter)
Dispersion Grating Coupling optics and laser filtering optics
Sampling optics (microscope objective, macro optics)
A little bit of history
Raman’s earliest spectra recorded with small prism spectroscope and photographic plate Excitation at first via filtered sunlight then via Hg arc lamp with filter to select Hg line of interest Sample (liquid) in large spherical flask
Collection times of hours to days
Very long sample preparation times
Advent of laser
Multi-stage spectrographs
Single-channel detectors
Structure of presentation
Follow “path” of light in diagram on previous slide Light source – usually laser Pre-sample filtering Collection optics
Confocal Raman microscope / spatial resolution Polarisation control Wavelength selector
Filtering of Rayleigh scattered light Collection of Raman photons Microscope and Macrochamber approaches
Spectral resolution and spectral coverage
Detector options “Hyphenated” Raman techniques
Raman spectroscopy at non-ambient conditions
Light source
Raman spectroscopy requires a source
With high intensity With monochromatic output
Best type of source is a laser Dispersive Raman instruments can be equipped with laser sources from UV to NIR
Possible wavelengths are (in nm): 227, 244, 257, 325, 364, 413, 442, 457, 473, 488, 514, 532, 633, 647, 660, 785, 830, 1064
Light source
Reasons to use different laser wavelengths:
Avoid or suppress fluorescence
Resonance Raman spectroscopy
A note on units used for Raman spectroscopy
Wavenumber (cm-1) is an energy unit
Raman peaks are always at the same relative position to the exciting laser line
Same cm-1 position, different nm position E.g. diamond Raman peak is at 1332 cm-1. In nm terms the diamond Raman peak is at 521.9 nm relative to a 488 nm laser line, or at 552.4 nm relative to a 514.5 nm laser line or at 876.5 nm relative to a 785 nm laser line Luminescence peaks have absolute energy values, so the nm position stays constant, but the cm-1 changes depending on the excitation wavelength E.g. PL peak at 575 nm is at 2044 cm-1 relative to 514.5 nm, but at 3101 cm-1 relative to 488 nm
Basic layout of dispersive Raman spectrometer Polarisation control Light source
Pre-sample filter
Sample Collection optics
NEXT
Post-sample filter Polarisation analyser Wavelength selector (monochromator or spectrograph)
Detector
Pre-sample filtering Optical filter between source and sample to “clean up” incoming beam.
60000
e.g. plasma lines from gas lasers
70
without line filter with line filter
50000
Intensity (arb. units)
70000
40000 30000 20000 10000
514.5nm line filter
0
50 400 40
80
30
70
600
800
1000
1200
1400
1600
1800
Wavenumber (cm-1)
60
20 10 0 400
500
600
Wavelength (nm)
Transmission (%)
Transmission (%)
60
50 40 30 20 700
800
10 0 500
505
510
515
520
Wavelength (nm)
525
530
Data measured by the author
Sample Solid, liquid and gas samples can be analysed
For liquids and gases, typically through sides of transparent container
For solid samples, range from large single crystals to nano-sized powder can be accommodated. Flat surface advantageous; not critical. Choice of microscope objective becomes important Rule of thumb – if you can get it under the microscope, you can try to measure it!
10
Graphite powder under Raman microscope with 100x objective
Length Y (µm)
20
30
40
50
60
70 0
20
40
60 Length X (µm)
80
100
Basic layout of dispersive Raman spectrometer Polarisation control Light source
Pre-sample filter
Sample Collection optics Post-sample filter
Red components = essential
Polarisation analyser Wavelength selector
NEXT
(monochromator or spectrograph)
Detector
Collection geometries
Collection geometry for Raman scattered light
900
Theoretically from any angle Practically and historically three angles have been favoured: 00, 900 and 1800 1800
00
Incident laser
Sample molecule
1800 Backscattering collection geometry
It is the configuration possible for all types of samples Most common collection geometry now due to wide use of microscope optics Implementation difficulty – common light path for incident and scattered light Laser is guided to sample by beam splitter OR injection-rejection filters (edge or notch filters) The filter reflects the laser beam wavelength but transmits Raman shifted wavelengths
Incident laser
Sample molecule
Filtering of Rayleigh scattered light
Raman instruments can be divided into two principal groups depending on the technique employed to filter out the Rayleigh scattered light from the Raman scattered light First group is single stage instruments that suppress the Rayleigh light by notch or edge filters Second group is double or triple stage instruments that suppress the Rayleigh light by an intermediate slit
Difference between notch and edge filters Comparison between white light transmission characteristics for a notch filter and an edge filter (example is for 514.5 nm)
90
90
80
80
70
70
60
60
Transmission (%)
Transmission (%)
514.5nm notch filter
50 40 30 20
514.5nm edge filter 50 40 30 20
10
10
0
0
400
450
500
550
600
Wavelength (nm)
650
700
750
400
450
500
550
600
650
700
750
Wavelength (nm)
Data measured by the author
Difference between notch and edge filters (2) Same data, but now in wavenumber units (cm-1)
90
90
80
80
70
70
60
60
Transmission (%)
Transmission (%)
50 40 30 20
40 30 20
10
10
0 -400
50
0 -300
-200
-100
0
100
200
300
400
Wavenumber (cm-1)
-400
-300
-200
-100
0
100
200
300
400
Wavenumber (cm-1)
Zero Raman shift = Excitation laser position Finite lifetime
Virtually infinite lifetime
Stokes and anti-Stokes Raman
Stokes Raman ONLY
Data measured by the author
Filters vs. subtractive double monochromator
Can also filter the Rayleigh scattered light via a double or triple spectrometer with an intermediate slit
Traditional way before availability of notch / edge filters
Cut-off with double or triple spectrometer is < 5 cm-1 (all λ)
900 Collection geometry
This configuration (also called right angle scattering) was initially used since Rayleigh scattered light was minimised. It is only useful for transparent samples (both solid and liquids) Often used in macro-Raman experiments OR where Raman selection rules are being investigated Incident laser beam path different from scattered light beam path
Raman scattered light Sample molecule Laser
00 Collection geometry – transmission Raman
00 collection geometry for Raman volume measurements of bulky opaque sample scattered light Often used for rapid, non-invasive and non-destructive analysis of pharmaceutical components of tables and capsules as it is highly representative of the bulk material
Laser
Raman scattered light Sample molecule
Collection optics – microscope objectives
Depending on the experiment and the sample, different microscope objectives are used to collect the sample signal Micro-Raman instruments are equipped with a range of objectives with different numerical aperture Objective
N.A.
Working distance (mm)
100x
0.90
0.21
50x
0.75
0.38
10x
0.25
10.6
100x LWD
0.80
3.4
50x LWD
0.50
10.6
Collection optics – Numerical Aperture (N.A.)
N.A. = n x sinθ n : refractive index
θ : aperture angle
Collection optics – choice of objective (1)
Choice of microscope objective depends on the transparency of the sample For transparent samples, low N.A. objective works better
Larger degree of penetration → larger sampling volume → increased signal
Raman spectra of LiNbO3 as a function of objective
Intensity (arb. units)
20x LWD 50x LWD 100x
100
Data measured by the author
200
300
400
500
600
Wavenumber (cm-1)
700
800
900
Collection optics – choice of objective (2) Choice of microscope objective depends on the transparency of the sample For opaque samples, high N.A. objective works better
Intensity (cnts/s)
Higher laser power density → increases sensitivity Wide collection angle → increased signal
Wavenumber (cm-1)
Confocal characteristics and axial resolution (1)
Axial spatial resolution (i.e. in “z”-direction) is determined by size of confocal pinhole in a confocal microscope Decreasing confocal pinhole diameter increases axial and lateral resolution
Diameter of order of tens of microns to hundreds of microns
} Incident laser beam
Sample
Confocal characteristics and axial resolution (2)
Confocal characteristics and axial resolution (3) From Barbillat et. al., J. Raman Spec. (1994) 25 3
Confocal characteristics and axial resolution (4)
Axial resolution as a function of confocal pinhole diameter for silicon using λexc= 633 nm
Narrow hole: Collects Raman light that originates only from within a diffraction limited laser focal volume with dimensions of: Focus waist diameter ~ 1.22 λ / NA Depth of laser focus ~ 4 λ / (NA)2
Example of effect of confocal pinhole diameter
Effect of refractive index on axial resolution
Spatial resolution, especially axial resolution, is improved if refractive indices of the media match closely (n1 ≈ n2 or ideally n1 = n2) (with water or oil immersion objectives) With normal (air) objectives n1 ≠ n2 (air and sample material) and optical distortion results which reduces spatial resolution
) Data and figures from paper by Neil J Everall, Applied Spectroscopy (2000) 54 773
Use of water immersion objective
Exploring lateral spatial resolution
Lateral spatial resolution determined by laser wavelength λ and numerical aperture (N.A.) of objective Laser spot diameter d given by
1.22λ d= NA Table of d (nm) for several NA and λ
Diffraction limited resolution d given by Rayleigh criterium
0.50
0.50
0.90
(UV 40x)
(Vis LWD 50x)
(Vis 100x)
595
-
-
514.5
-
1255
697
785
-
1915
1064
λ(nm)
244
NA
Lateral spatial resolution example (1)
Circles etched by Ga ion beam on silicon at different distances from each other (1400, 1200, 1000, 800, 600, 400, 200 nm)
Data courtesy of Dr P Wilhelm, TU Graz
Lateral spatial resolution example (2)
Circles were mapped with a mechanical table and 100x dry objective with NA = 0.9 at λexc = 633nm
Localised heating with microscope objectives
Microscope objective focuses laser beam down to small spot High resulting power densities can heat or burn sample
For 100x objective with N.A. = 0.90, diameter ~700 nm with 514.5nm laser If laser power at sample is 1 mW, then power density is 260 kW/cm2 If laser power at sample is 5 mW, then power density is 1.3 MW/cm2
Mitigate by turning down laser power, slight defocusing, smaller NA objective, or by scanning laser spot over set area (e.g. 10x10µm2) – much lower power density Laser damage (514.5nm) in graphene oxide even at low power density of 640 W/cm2. 5µm
No burning, acceptable signal with 64 W/cm2. Data measured by the author
Collection optics – remote probes
Instead of microscope or macrochamber, can have remote probe linked to input port of spectrograph via fibre optic cables Ideal for remote Raman measurements such as e.g. in situ, noninvasive chemical analysis Can have versions suitable for immersion, or high temperature and/or pressure. Application to solids, liquids and gases, from research to process control Hand-held Raman spectrometers
Basic layout of dispersive Raman spectrometer Polarisation control Light source
Pre-sample filter
Sample Collection optics Post-sample filter
Red components = essential
Polarisation analyser Wavelength selector
NEXT
(monochromator or spectrograph)
Detector
Polarisation control
As seen in section on Raman theory, the polarisation of the Raman scattered light can contain useful information This can be measured by using polarised excitation (laser source) and a polarisation analyser Laser light is usually already plane polarised, so a polariser in the excitation path is not necessary Can rotate the direction of the plane polarisation of the laser using a λ/2 wave plate (some wavelength dependence) OR a double Fresnel Rhomb prism (virtually wavelength independent) In the Raman scattered beam path an analyser can be set both perpendicular and parallel to the incident plane of polarisation
Polariser Analyser
Polarisation control – position of elements Pre-sample filter
Polarised light source
Sample Collection optics
λ/2 wave plate to turn the polarisation of the polarised laser
Post-sample filter Analyser perpendicular or parallel to the excitation plane Wavelength selector (monochromator or spectrograph)
Detector
Polarisation measurements
Polarisation measurements: Isotropic sample Cyclohexane
Polarisation measurements: Anisotropic sample Acetylic salicylic acid
Basic layout of dispersive Raman spectrometer Polarisation control Light source
Pre-sample filter
Sample Collection optics Post-sample filter
Red components = essential
Polarisation analyser Wavelength selector
NEXT
(monochromator or spectrograph)
Detector
Spectral resolution and spectral coverage
What do we mean by spectral resolution?
Spectral resolution is a function of
Dispersion Entrance slit width(s) Pixel size of CCD detector
Dispersion is a function of
It is the capability to resolve peaks For sharper and/or closely spaced peaks, require higher spectral resolution
Focal length of spectrograph Groove density of the grating The excitation wavelength
In general, a long focal length spectrograph and a high groove density grating give the best spectral resolution Differences between mirror-based optics and lens-based optics
What does a spectrograph look like?
Dispersion as a function of excitation wavelength
Spectral coverage – function of λ
Dispersion as a function of the grating (1) Grating Equation:
sin i + sin i’ = nkλ Where n is the groove density, k is the diffraction order and λ is the wavelength
Linear dispersion:
dx Fkn = dλ sin i'
Where F is the focal length of the spectrograph Reciprocal linear dispersion:
dλ sin i ' = dx Fkn
Dispersion as a function of the grating (2)
Dispersion as a function of the grating (3) Example of graphite measured with 514.5 nm excitation and 50 µm slitwidth for two different gratings
600 grooves/mm 1800 grooves/mm Intensity (arb. units)
1000
1500
2000
2500
3000
Wavenumber (cm-1) Data measured by the author
Dispersion as a function of focal length (1)
dλ sin i ' = dx Fkn
Dispersion as a function of focal length (2)
Spectral resolution as a function of slit width (1)
Entrance slit width is one of parameters determining spectral resolution The narrower the slit, the narrower the FWHM of a peak, and the higher the spectral resolution
When measuring a Raman line with natural width smaller than the spectrograph’s resolution, then the measured width will reflect the spectrograph’s resolution.
Spectral resolution as a function of slit width (2)
The narrower the slit, the narrower the FWHM of a peak, and the higher the spectral resolution Example of 546.074 nm green emission line from a Hg lamp 50 µm 100 µm 200 µm
Intensity (arb. units)
1110
1115
1120
1125
1130
1135
Data measured by the author
Wavenumber (cm-1)
Hg lamp 546.074 nm emission line as a function of entrance slit width
Spectral resolution as a function of CCD pixel size
CCD detector is made up of very small pixels, so each pixel serves as an exit slit to the spectrograph For CCDs with same physical size, the CCD with a larger number of (smaller) pixels produces a larger number of spectral points closer to each other thus increasing the limiting spectral resolution and the sampling frequency
26 µm pixel vs 52 µm pixel (simulation)
Basic layout of dispersive Raman spectrometer Polarisation control Light source
Pre-sample filter
Sample Collection optics Post-sample filter
Red components = essential
Polarisation analyser Wavelength selector (monochromator or spectrograph)
NEXT
Detector
What types of detector are available?
Single-channel detectors
Multi-channel detectors
Traditionally Photomultiplier Tube (PMT) Used in conjunction with scanning multi-stage monochromators First “multi-channel” detector was photographic plate! Charge-Coupled Device (CCD) detectors commercially available since early 1990s. An array detector. Continuous development Also Intensified Photodiode Array, Charge-injection devices, nonsilicon detectors
CCD detectors
Close to ideal detector for Raman spectroscopy Based on Silicon, spectral range from 200nm to 1100nm, high QE Array of pixels, 10 – 20 µm square Array size 1024 x 256 or 2048 x 256 Y Cooled for low electronic noise, low dark current Cosmic Ray Events (CREs) ∆ν
What does a CRE look like?
CREs result from random background radiation caused by radioactive decay of elements and random cosmic events Visible as random, very sharp, narrow peaks, often only one or two pixels wide Removed by software correction graphite spectrum with CREs graphite spectrum without CREs Intensity (arb. units)
1200
1300
1400
1500
1600
Wavenumber (cm-1)
1700
1800
Data measured by the author
CCD detectors – sensitivity and spectral range
Comparison of different detectors regarding their quantum efficiency between 200 nm and 1100 nm
CCD detectors – effect of sensitivity in NIR (1)
Comparison of fingerprint regions of polyethylene spectra CCD quantum efficiency and Raman scatter efficiency (~ ν4) has influence on sensitivity in NIR
Factor of 6 in measurement time, factor of 2 in laser power → “12” times more input in NIR
CCD detectors – effect of sensitivity in NIR (2)
Comparison of fingerprint regions of polyethylene spectra CCD quantum efficiency and Raman scatter efficiency (~ ν4) has influence on sensitivity in NIR, especially at longer wavelengths
Signal 25 times stronger with 633 nm laser in spite of “12” times more input in NIR
CCD detectors – effect of sensitivity in NIR (3)
CCD quantum efficiency and Raman spectra of polyethylene compared directly on one graph. Lower QE of detector in NIR plays important role (note wavelength, not wavenumber scale)
Light flux as a function of dispersion (1)
If dispersion is doubled, then integrated signal becomes approximately halved since every pixel is collecting only half the number of photons
Light flux as a function of dispersion (2)
Intensity (cnts/sec)
Silicon spectra acquired with 633 nm with 600 and 1200 grooves/mm gratings. Ratio is 1 : 2 Dispersion is 2.07 and 0.94 cm-1/pixel @ 654 nm (Si Raman peak in nm units with 633nm excitation. Ratio is 2.2 : 1 Integrated Raman signal of Si Raman peak is 265063 and 136982 Ratio is 1.9 : 1
Intensity (cnts/sec)
Light flux as a function of dispersion (3)
If dispersion is similar or equal by a combination of grating and focal length (e.g. F = 300 mm & 1200 gr/mm and F = 600 mm & 600 gr/mm) the integrated signal is approximately the same
dλ sin i ' = dx Fkn
“Hyphenated” and other Raman techniques
“Hyphenated” techniques refer to Raman spectroscopy combined with other analytical techniques, such as
Can also do Raman under non-ambient conditions using various stages and cells
Raman-AFM Raman-TERS Raman-SEM Raman-FTIR
High temperature (up to 1500oC) in microscope furnace, e.g. Linkam stages Low temperature (77K or 4K) in microscope stage or cryostat, e.g. Linkam stages and Oxford Instruments cryostats High pressure measurements (several GPa to 100 GPa) in Gem Anvil Cells (GACs) and Diamond Anvil Cells (DACs)
Other techniques not touched on
SERS – Surface-enhanced Raman scattering CARS – Coherent Anti-Stokes Raman scattering Hyper Raman
Summary and conclusions
Overview of elements of instrumentation necessary for successful Raman spectroscopy Many aspects only touched on briefly With modern instrumentation relatively easy to get Raman spectrum, but optimisation requires some patience and experience Several textbooks available with more detailed discussions
Acknowledgments
Part of this presentation is modelled on that of Dr Bernd Bleisteiner of HORIBA Scientific. Use of this material is gratefully acknowledged. Several examples were from data measured at Wits University. Permission to use this data is gratefully acknowledged.
Thank you for your attention!