Professional Documents
Culture Documents
h
Raman spectroscopy Dt DE (1)
2
When light of a certain wavelength interacts with
a molecule, most photons are elastically scattered and With typical photon energies of 1–2 eV, the lifetime of the
therefore have the same energy as the incident photons. excited state is only about 10–15 s. After this extremely
However, a very small fraction (approximately 1 in 106– short time, the molecule falls back either to the vibrational
107 photons) is in elastically scattered, which means that ground state or to an excited state (Figure 1). When the ini-
the energy of the scattered photon is different than the tial and final states are identical, the process is called Ray-
energy of the incident photon. leigh scattering.
This is called the “Raman effect”, and it was discovered If the initial state is the ground and the final state
by Sir Chandrasekhara Raman in 1928 (Raman, 1928; a higher vibrational level, the process is called Stokes scat-
Raman and Krishnan, 1928). Unlike today, he used tering, if the initial state is energetically higher than the
a filtered beam of sunlight as an excitation source and final state, this is referred to as Anti-Stokes scattering.
his eye as a detector for the frequency shifted light. This The difference in energy between the incident and the
was long before the development of the first laser by Raman scattered photon is equal to the energy of
Maiman in 1960. Raman was awarded the Nobel Prize a vibration quantum of the scattering molecule. A plot of
in 1930 for this discovery. The theory behind the Raman intensity of scattered light versus energy difference is
effect was derived five years earlier by Smekal (1923). called a Raman spectrum.
The tremendous importance of the Raman effect lies in The position of a Raman line is usually given in relative
the fact that the energy shift between the exciting and wavenumbers (1/cm), which is the energy shift relative to
the Raman scattered photon is caused by the excitation the excitation line:
(or annihilation) of a molecular vibration. This energy
shift is thus characteristic and therefore a fingerprint for 1 1
the type and coordination of the molecules involved in n¼ (2)
the scattering process. lincident lscattered
Raman Microscopy (Confocal), Figure 1 Energy level diagram for Raman scattering.
756 RAMAN MICROSCOPY (CONFOCAL)
Raman Microscopy (Confocal), Figure 4 Collection efficiency Raman Microscopy (Confocal), Figure 5 The intensity
as a function of the pinhole size normalized to the total power in distribution of two point sources which are separated by the
the image plane. Rayleigh criterion.
758 RAMAN MICROSCOPY (CONFOCAL)
changes would be to increase the laser power and to Czerny-Turner spectrometers only have a transmission
increase the integration time. However, there are limita- of about 30% at this wavelength.
tions to these possibilities as shown below: Efficiency of the detector (CCD): Back-illuminated CCD
(a) Laser power: If a power of only 1 mW from a 532 nm cameras show a quantum efficiency (QE) of >90% over
laser is focused diffraction limited using a 100 air the entire spectral range of interest for 532 nm excitation.
objective with a NA of 0.9, the power density in the Deep depletion back-illuminated CCD cameras, on the
illuminated spot is approximately 106 W/cm². These other hand, show a QE of >90% for 785 nm excitation.
levels are only possible because a single point is illu- As a comparison, front-illuminated CCD cameras do not
minated and the heat effectively dissipates in three exceed 55% quantum efficiency at any wavelength.
dimensions. Nevertheless, the maximum laser power Additionally, the dark current of the cameras needs to
on the sample is therefore typically in the range of be minimized, which is achieved through efficient Peltier
1–20 mW. cooling.
(b) Integration time: Increasing the integration time will The readout noise is another limiting factor for small
increase the signal to noise ratio significantly. How- signals. As the analog to digital (A/D) converter of all
ever, when CRM or mapping is applied, in which an cameras will add at least 5–10 electrons read-out noise
image is produced by recording a spectrum at every to the signal, any signal below approximately 5 electrons
image pixel, the integration time per spectrum needs will be lost in the noise. Additionally, the faster the A/D
to be kept to a minimum. An integration time of only converter is operating, the higher the read-out noise
1 s per spectrum will, when recording an image of will be. Electron-multiplying CCD (EM-CCD) cameras
128 128 pixels, result in a total acquisition time can be used to overcome this problem. With these
of about 4.5 h, which is an inconvenient time-span cameras, the signal is amplified before the A/D conver-
for routine application. sion, allowing the detection of even single photons
and reducing the necessary integration time down to
Therefore, a system for CRM must be capable of obtaining milliseconds.
the Raman spectra with an acceptable signal-to-noise
ration in less than 50–100 ms. There are several parts of Principle of operation
a confocal Raman microscope which should be optimized Confocal Raman microscopes generally provide a variety
in order to allow such rapid data acquisition. These will be of modes of operations. The most common are listed
discussed in the following: below:
Laser power: As described above, the maximum laser Collection of Raman spectra at selected sample areas
power is limited and will heavily depend on the destruc- (single spectrum): Single Raman spectra can be collected
tion and alteration threshold of the sample used. at user-selectable sample areas with integration times
Collection efficiency of the objective: Using objectives ranging from ms to hours. The position of the collected
with high numerical apertures maximizes the collection spectrum can normally be fully controlled in 3D.
efficiency since light from a larger solid angle is collected. A stable and precise positioning system must be included
Throughput of the microscope: In order to enhance the in the instrument to ensure that the point of interest will
throughput of the microscope, the components used remain fixed under the excitation focus. This is very
within the beam path should be optimized and if possible important when spectra with longer integration times
minimized since every passive optical element introduces for the best quality and signal to noise ratio are to be
losses to the signal. obtained from extremely small sample volumes. For
Efficiency of the grating: Highly efficient gratings with example, using an oil immersion objective (NA 1.4) with
the correct blazing angle for the excitation wavelength a 532 nm laser and the proper pinhole size allows the sam-
should be used. As an example: a grating blazed at ple volume to be as small as 230 230 550 nm.
500 nm will have an absolute efficiency of >60% up to Collection of time series of Raman spectra at selected
3000 rel.1/cm when using a 532 nm laser. Using the same sample areas (time spectrum): With this mode, time series
grating with a 785 nm laser will cause the efficiency to of Raman spectra can be obtained to analyze dynamic
drop to about 30% at 3000 rel.1/cm. The correct grating sample properties. Thousands of spectra can be obtained
blazed at 750 nm, however, shows an efficiency >60% over time and analyzed with integration times ranging
up to 3000 rel.1/cm when used with the 785 nm laser. from ms to tens of seconds.
Efficiency of the spectrometer: There are a variety of spec- Raman spectral imaging: In the Raman spectral imaging
trometers available on the market. While mirror-based mode, the sample is moved in X and Y and a full Raman
spectrometers can generally be used over a wide spectral spectrum is obtained at every pixel measured. From these
range, lens-based spectrometers often show a higher data sets images of, e.g., the integrated intensity of various
throughput in the spectral range they are designed for. bands can be generated. This is illustrated in the following
Some lens-based spectrometers show a transmission at example.
532 nm of >60%. Spectrometers using a combination of A drill core section from the Äspö Hard Rock Labora-
lenses and prisms or mirrors generaly have a significantly tory, Sweden, was studied using the large area scan mode
lower throughput. Commercially available mirror-based CRM, aiming to characterize secondary fracture fillings in
RAMAN MICROSCOPY (CONFOCAL) 759
a 1.8–1.4 billion years old diorite. The rock sample was Other spectra of the same scan show the characteristic fea-
obtained from 450 m below the surface (kindly provided tures of calcite or fluorite (Figure 6d). Using these, the cal-
by SKB, Swedish nuclear fuel and waste management). cite (Figure 6b) and fluorite (Figure 6c) images can be
This sample was examined with a 532 nm laser and generated by using additional integral filters for the
a 50 air objective (NA 0.55). The scan range was marked regions. Other features of the spectra such as the
8000 μm in X and 2000 μm in Y with 800 200 points width of peaks or their position can also be evaluated by
resolution. Each spectrum was integrated for 36 ms. applying the corresponding filters.
Figure 6 shows the characteristic spectra of calcite, fluorite Further evaluation of the data allows the averaging of
and quartz found in the sample. Integrating over, for similar spectra (for which a cluster analysis is often used)
example, the area marked in green in Figure 6d results in and the subtraction of, for example, pure spectra from
a single value for each of the 160,000 spectra. These spec- mixed spectra to extract the spectra of the various compo-
tra can be displayed as a color-coded image as shown in nents (spectral de-mixing). These spectra can then be used
Figure 6a for quartz. Here, brighter values indicate with the basis analysis, where each of the spectra recorded
a higher integrated intensity of the quartz peak and the dis- are fitted with a linear combination of the basis spectra.
tribution of quartz can thus be obtained from this image. The result of such an analysis is one image for each basis
Raman Microscopy (Confocal), Figure 6 Spectra (d) and spectral images of quartz (a), calcite (b) and fluorite (c) recorded from the
diorite (a) and the fracture fillings (b and c). The data were recorded with a WITec alpha500R confocal Raman microscope.
Raman Microscopy (Confocal), Figure 7 DE-mixed and averaged spectra of the various components in the diorite and the adjacent
fracture minerals with the combined image on the left.
760 RAMAN MICROSCOPY (CONFOCAL)
Raman Microscopy (Confocal), Figure 8 Variations in the quartz phases due to changes in the crystallographic orientation, the
crystallinity, and the minerals.
Cross-references
Biosignatures in Rocks
Terrestrial Deep Biosphere
REDUCTION SPHEROIDS
Beda A. Hofmann
Natural History Museum Bern, Bern, Switzerland
Synonyms
Bleaching haloes; Fish eyes; Radioactive concretions; Reduction Spheroids, Figure 1 Reduction spheroids in Triassic
Radioactive nodules; Reduction mottling; Reduction red bed (marly dolomite), Chinle Formation, Paradox Valley,
spots; Reduktionshöfe Colorado, USA. These reduction spheroids are strongly enriched
in V, Cr, and Au. Diameter of larger reduction spheroid is 8 cm.
Definition
Reduction spheroids are small-scale (diameter: few milli- Pb-depleted haloes indicates late diagenetic ages
meters to maximum about 20 cm) spheroidal bleached (Hofmann, 1990; Hofmann and Frei, 1996). Reduction
features in red-bed sediments characterized by the absence spheroids often occur in large numbers in red beds and
of hematite and often containing a mineralized core can constitute up to several vol% of the rock. Based on
(center, Figure 1). Reduction spheroids result from local textural evidence, such as fracture-bound occurrences
chemical reduction of ferric iron and other elements. Host and mass-balance calculations (Hofmann, 1991), reduc-
rocks are marine and continental red beds and, more tion spheroids cannot be due to detrital organic-rich parti-
rarely, altered, hematite-stained crystalline rocks. cles. Instead, the formation must be due to locally
Reduction spheroids have been described from many catalyzed redox reactions between a mobile reductant
occurrences in red beds of widely variable age worldwide: and oxidants. The presence of isotopically light sulfides
Proterozoic of Canada (Tanton, 1948); Cambro-Silurian in some cores (Hofmann, 1991), indicating diagenetic sul-
marine red beds of the NE USA (Beutner and Charles, fate reduction, as well as the need for a local catalyst,
1985); Devonian continental red beds of Scotland strongly favors a microbiological origin of reduction
(Parnell, 1985); Carboniferous continental red beds of spheroids. Reductants may be mobile hydrocarbons or
New Brunswick and Nova Scotia, Canada (Dyck and molecular hydrogen resulting from the radiolysis of pore
McCorkell, 1983); Permian continental red beds of Ger- water (Hofmann, 1992). The formation of reduction spots
many (Eichhoff and Reineck, 1952; Mempel, 1960; is most likely due to chemolithotrophic communities.
Schreiter, 1930), Switzerland (Hofmann, 1990, 1991), Kerogen-like organic matter is occasionally present at
Southwest England (Harrison, 1970; Perutz, 1940), and some localities (Curiale et al., 1983; Parnell, 1985) but is
Cretaceous radiolarian cherts of Oman (Hofmann, 1991). lacking at many other occurrences (Hofmann, 1993).
Reduction spheroids characteristically consist of
a spheroidal volume of rock from which hematite pigment
has been dissolved. Central dark cores often display pro- Conclusion
nounced concentric zonation. The cores are strongly Reduction spheroids are very common in red-bed
enriched in numerous redox-sensitive elements (dominant sediments, and there are strong indications for an origin
V, but also U, Se, As, Ni, Co, Cu, Ag, Au, PGE, and due to redox reactions between a mobile reductant
others). The dominant mineral in reduction spheroid cores and oxidized elements locally catalyzed by microbial
is the vanadium mica roscoelite KV2AlSi3O10(OH)2. Ura- activity. Reduction spheroids are thus easily visible
ninite is another very common mineral, leading to an often features resulting from redox processes catalyzed by
increased radioactivity of reduction spheroids (Figure 2). a deep biosphere of sediments, and represent a type of
Arsenides of Ni, Co, and Cu are also quite com- biosignature that may also be present in oxidized rocks
mon. Radiometric dating of roscoelite, uraninite, and on Mars. Reduction spheroids contain small-scale