Abstract

Raman spectroscopy is one of the most informative probes for studies of material properties under extreme conditions of high pressure. The Raman techniques have become more versatile over the last decades as a new generation of optical filters and multichannel detectors become available. Here, recent progress in the Raman techniques for high-pressure research and its applications in numerous scientific disciplines including physics and chemistry of materials under extremes, earth and planetary science, new materials synthesis, and high-pressure metrology will be discussed.

1. Introduction

Raman spectroscopy probes elementary excitations in materials by utilizing inelastic scattering processes of a near ultraviolet, visible, and near infrared monochromatic light source (commonly a laser). The scattered radiation forms a spectrum near that of the excitation laser wavelength. The results are easy accessible for recording as very sensitive detectors are available in this spectral range.

The main strength of the Raman spectroscopy is the ability to provide a great wealth of easily analyzable information very rapidly. The Raman spectra can be used to characterize the elastic, vibrational, electronic, and magnetic subsystems through the observations of the corresponding elementary excitations. The Raman spectra of phonons (lattice and molecular vibrations) have a very high selectivity, thus permitting finger-printing analysis of the materials phase that can include its composition and state. The information about the crystal structure is provided through the vibrational selection rules, which govern the Raman activity of phonon modes depending on their symmetry and the wave vector. The access to electronic and magnetic states is attained through the coupling to the vibrational states, and through the observations of the spectra of the electronic and magnetic excitations.

As the spectra of elemental excitation change with the application of pressure, the associated phenomena can be effectively studied by in situ Raman spectroscopy. These include changes in the energy of the vibrational excitations, phase transformations (including melting), chemical reactivity, and magnetic and electronic transitions. This information can be used for multiple applications because these transformations are of interest for fundamental physics and chemistry, materials, and earth and planetary science. Last (but not least), Raman spectroscopy is a great tool for measurements of pressure at both extremely low and extremely high temperatures.

Previous comprehensive reviews on Raman spectroscopy at high pressures [13] have mainly discussed mineralogical and environmental applications. Other reviews [46] need updating for recently developed techniques and newly obtained groundbreaking experimental results, which often created new scientific understanding. Moreover, a field of high-pressure high-temperature Raman studies using laser heating in diamond anvil cells (DACs) has only recently become mature. All this made a new review paper desirable. Here, the author presents a review of Raman spectroscopy methods and applications at high pressures, which include these new developments.

2. Development of Raman Techniques

In high-pressure Raman instrumentation, special care must be taken to interface the technique with the DAC. The typical samples dimensions in the DAC are several tens of micrometers and a few micrometers in thickness; these dimensions are even smaller if measurements to ultrahigh pressures (>100 GPa) are planned. The presence of an optically thick diamond window (≥2 mm) in the optical path causes large geometric and chromatic aberrations, which affect the image quality, the signal collection efficiency, and the depth selectivity. Until recently the use of optics with a large numerical aperture was not feasible because it conflicts with the requirements of stability of diamond anvils and supporting backing plates. Objective lenses with a large working distance (normally ≥10 mm) are also necessary. Moreover, diamond windows in the optical path produce parasitic back reflections, which make Raman measurements at low frequencies difficult. This is because the specularly reflected and elastically scattered laser light increases the background level thereby decreasing the signal-to-noise ratio. Also, diamond anvils fluorescence in the laser beam also contributes to the background level. All these factors call for the custom-made laser microscopes to interface the DAC as the commercial microscopes typically do not provide a sufficient level of flexibility to address all the issues described above.

The custom Raman microscope for DAC has been described in our previous publications [57] (Figure 1). Here the author will focus on the presence of the intermediate field aperture (spatial filter) to reduce the spurious background, the ability to easily change the excitation wavelength (as this helps to provide the optimum excitations conditions and to decrease the fluorescence), and the flexibility in choosing the scattering geometry.

A confocal field aperture with a diameter corresponding to the dimension of the laser spot attenuates the spurious elastically scattered laser radiation substantially. Moreover, it makes the depth of focus smaller thus cutting substantially the fluorescence background from the diamonds. Although the intermediate field apertures are redundant in the sense that the confocal geometry can be realized using the entrance slit of the spectrometer and by narrowing the detector readout in the nonspectral direction [8], the latter strategy is insufficient for the DAC operation (based on our experience) because of a substantially increased level of the spurious radiation compared to that for free-standing samples.

The use of angular excitation geometry (Figure 1) allows a drastic reduction in background spurious radiation compared to pure backscattering geometry, thus increasing the signal-to-noise ratio and making easier the access to the low-frequency spectra. Moreover, this technique allows reducing substantially Raman and fluorescence background signals from the diamond anvils because the corresponding signals, excited mainly off axially, are suppressed more efficiently than in the case of the axial laser illumination (pure backscattering geometry). This will be illustrated later on examples of studies of metals and high-temperature superconductors. A conventional backscattering geometry can also be used for the system presented in Figure 1 by introducing a beamsplitter (neutral optical density or dichroic).

To realize the angular excitation geometry the DAC must be designed or modified to allow the angular light access. Previously, specially designed tungsten carbide seats with angular conical holes have been used for this purpose [5, 9]. To focus the laser radiation on the sample, one needs to use a separate lens (or an objective lens). Due to geometric considerations, this lens has longer focal length compared to that which collects the scattered radiation. This effectively increases the minimum laser spot dimension at the sample. In addition, use of angular laser incidence geometry results in further increase of the laser spot because as off-axis geometric aberrations (e.g., astigmatism) which affect the dimensions and the shape of the focused laser beam in this case. Both these factors place limits on the applicability of this geometry (e.g., for heterogeneous samples). Nevertheless, the use of the angular geometry was found to be crucial for obtaining high-quality spectra in the case of metals [5, 6, 10, 11].

Raman spectroscopy has improved substantially in its ability to perform rapid measurements due to the development of extremely sensitive and low-background array detectors (CCD), holographic transmission optics [12], and fast imaging spectrometers (e.g., [1, 13]). More recently, new hard-coated optical filters [14], which have comparable transition width of the transmission band to those manufactured using a holographic technology, have appeared on the market. Notch, bandpass, and dichroic filters/beamsplitters using this technology efficiently block undesirable Rayleigh-scattered light and obviate the need for cumbersome low-transmission double subtractive monochrometer-filter systems for routine measurements above 50 cm−1. Moreover, recently a new generation of notch and bandpass filters with a transition width as narrow as 10 cm−1 has been developed based on a solid-state technology [15]. These filters are expected to be much more environmentally stable, and show no degradation over time.

Application of Raman spectroscopy for studies at very high pressures often requires capability to perform a rapid change of the excitation wavelength. This is needed to determine the nature of the bands observed (e.g., Raman versus fluorescence), to provide the optimum excitation (e.g., for ruby fluorescence), and to study the resonance phenomena (e.g., [16]). Modern Raman spectrometers, which use the very light-efficient and narrow transition width filter technologies (see above), are less flexible for tuning of the excitation wavelength compared to conventional triple-stage spectrometers. Nevertheless, the micro-Raman system, which is described here (Figure 1) is capable for a rapid change in the excitation wavelength covering the spectral range up to 100 nm without the need to use another set of holographic filters. This is achieved by changing the angular positions of the filters with respect to the laser and the scattered beam directions. These rotations of the filters do not affect the system alignment.

One essential improvement in the Raman techniques for high pressure would be a confocal Raman imaging (e.g., in the DAC). Newly designed laterally supported diamond anvils [17] allow using objective lenses with much larger numerical aperture (e.g., ) than with the anvils of conventional design. In principle, this makes 3D Raman mapping possible. A remaining obstacle is large optical aberrations introduced by diamond windows. These aberrations distort the wave front thus making the depth-selective measurements difficult. This problem can be solved using a specialized compensating optics; the work is currently in progress.

In situ Raman measurements in laser-heated DAC has been developed recently [18, 19]. Use of near IR lasers (1060 nm) as a heating source allows reducing the heating spot and producing more stable heating conditions compared to CO2 laser heating used previously [20]. These advantages have become even more noticeable when near IR fiber lasers with the improved mode quality and stability became commercially available. This improved laser beam quality allows very local heating of the samples in the DAC cavity, thus opening a possibility to use the technique at very high pressures. Recently, the technique has been further improved by the development of laser beam shapers, which allow controlling the laser power distribution in a focal plane [21, 22]. The use of a flat top or bimodal power distribution makes the heating conditions more stable and reduces the radial temperature gradients.

Heating of materials, that do not absorb the laser radiation is challenging. A thin foil with a small hole(s) or powder made of chemically stable metals (e.g., Pt) can serve as a laser beam absorber (coupler) thereby producing high-temperature conditions for the surrounding sample material [18, 19]. Arguably the best results for the in situ Raman spectroscopy have been obtained when couplers with small holes have been used [18, 19]. The coupler is heated by an IR heating laser spot, which is larger than the hole, while Raman scattering of the material in the hole is excited by another tightly focused visible laser. This geometry decreases the temperature gradients through the probed material. Furthermore, application of thermal insulation layers near the diamond anvil tips (which remain cool during the experiment owing to a very large thermal conductivity of diamond) eliminates the sample from these areas thereby further reducing the temperature gradients across the probed sample. A double-sided laser heating helps decreasing the temperature gradients across the sample contained in the coupler hole. The laser power can be split to two sides and controlled independently at each side using a system of polarization cubes and wave plates [22]. This system has been recently automated for data acquisition, which includes performing all the spectral measurements (Raman and radiative temperature measurements) at variable heating laser power [23]. This upgrade allows saving up to 90% time compared to manual data acquisition. This improvement is important for experiments with highly chemically reactive materials such as, for example, hydrogen [24].

The thermal radiation emitted by the sample and the coupler makes Raman measurements at high temperature challenging. One can avoid a substantial part of the thermal radiation by discriminating it spectrally and spatially. This can be achieved by using a blue excitation wavelength (e.g., 458 nm of a solid-state laser) and by using a confocal backscattering geometry. High-quality Raman spectra can be obtained up to approximately 2000 K in the experimental environment [19] (see also [25]). At higher temperatures, when thermal radiation rises above a critical threshold and obscures the Raman signal, a pulsed Raman system with 532 nm excitation synchronized with the gated intensified CCD detector can be used [26] (see also [27]) in combination with continuous [28] or pulsed laser heating [22]. As the detector collects the entire Raman signal (the Raman processes are very short, normally <1 ps) during the relatively short gate time and is closed between pulses, any continuous (or lasting longer) spurious radiation is suppressed by a factor that is proportional to the ratio of the time interval between pulses to the gate width. The suppression factors up to 50000 have been reported [26] providing the possibility of acquiring Raman signals at temperatures exceeding 5000 K.

The advances in Raman instrumentation introduced above have enabled a number of new opportunities for the Raman studies at high pressures, which will be described below. Arguably, the most prominent achievement is the ability of the modern Raman systems to perform very rapid measurements (<1 s acquisition time) thus enabling experiments under extremely high pressures and temperatures ( ).

3. Raman Studies at High Pressures

3.1. Phase Diagrams of Diatomic Molecules

Raman spectroscopy is ideally suited for studying the evolution of the bond character in diatomic molecules under compression because measurements of their vibrational properties provide a powerful diagnostic of this process. The electronic density function changes with compression to minimize the total energy of the system. For the molecular materials, pressure tends to destabilize the intramolecular bonds as the kinetic energy of the participating electrons increases steeply with compression (e.g., [29]). This process may be complicated by other phenomena such as, for example, the formation of some intermediate states [30, 31], which affect the molecular stability. Study of simple molecules such as diatomics is a valuable approach to solve this problem.

The behavior of hydrogen under pressure is of special interest because of the simplicity of the electronic structure and because of its dominant abundance in the universe, making it the major constituent of stars and planets. Today, Raman spectroscopy technique is still limited in pressure and temperature (to roughly below 150 GPa and 2000 K) making study of hydrogen under conditions of molecular dissociation very challenging. Therefore, the author will begin by reviewing the behavior of the halogens, which can be considered as analogous to hydrogen in electronic structure because hydrogen can be viewed as a group VII element [32], and the phenomena of interest in these materials occur at much lower pressures than in hydrogen (<100 GPa). Recently, iodine (I2) and bromine (Br2) have been extensively studied [33]. The application of Raman spectroscopy was crucial in revealing the bonding character change, the insulator-metal transition, and the presence of an intermediate charge modulated phase as a function of pressure [34]. Indeed, the Raman spectra revealed a number of changes with pressure, which can be associated with the observations obtained by other techniques, such as X-ray diffraction [34], X-ray absorption spectroscopy [35], and electrical conductivity [36]. Raman measurements show discontinuous frequency changes and the appearance of new bands that can be related to metallization. These observations could also be associated with an additional subtle phase transformations suggested by theoretical calculations [37], or both phenomena could occur simultaneously. Raman spectra show drastic changes when approaching the molecular dissociation transition. Just before the transition to the monatomic phase, a new low-frequency soft mode appears associated with the incommensurate structural and electronic modulation. The behavior of I2 and Br2 is very similar; moreover, the vibrational frequencies can be scaled suggesting the generality of the observed successive phase transition mechanism for the molecular dissociation [33].

The example of nitrogen reveals more complexity in high-pressure behavior, because in this case there is an additional possibility for creating the material with covalent bonding—single bonded polymeric phase [38]. The additional complexity is also added by extraordinary rich polymorphic behavior of molecular nitrogen revealed by Raman spectroscopy [3943]. The triply bonded N2 molecule is expected to destabilize at pressure above 50 GPa [38, 44], but Raman and infrared absorption (IR) experiments show its tenacious stability up to 170 GPa at 300 K (200 GPa at 80 K) [45, 46]. This behavior is because of kinetic boundaries between different classes of molecular and nonmolecular nitrogen phases [38]. Raman spectroscopy is a powerful technique to study numerous solid-solid phase transitions and metastability phenomena in nitrogen (and other molecular materials, see below) as it is very sensitive to minor changes in the crystal structure. This is because of the sensitivity of the Raman selection rules to the presence of atomic sites with different symmetry, and also to the vibrational coupling and the orientational ordering [40, 47]. Finally, Raman spectroscopy data (along with X-ray diffraction) were revealing in identifying the monatomic polymeric phase-cg-nitrogen. This transformation was found to occur abruptly above 110 GPa based on combined data of different research [28, 41, 48, 49]. This peculiar pressure dependence confirms that molecular dissociation is a volume-driven process that results in destabilization of the molecular orbitals. The associated with the molecular dissociation fluid-fluid transition was reported based on the results of shock-wave experiments [50] and theoretical calculations [51, 52]; this phenomena need to be studied in static high-pressure experiments to provide better insight into their nature as static experiments can allow more informative diagnostic tools. This topic will pose a challenge for the classical Raman spectroscopy tools in the years to come.

Nitrogen is also an excellent example of usefulness of Raman spectroscopy for determination of the melting line. The basis of this is a change in the vibrational selection rules in a fluid state. Indeed, one can expect that the phonon modes (translational intermolecular vibrations) should disappear in the fluid molecular phase because of a lack of translational order of the molecular centers of gravity. However, these observations can be confused with phenomena that accompany the transformation to an orientationally disordered solid phase (plastic phase). Fortunately, in the case of nitrogen there are concomitant with the melting changes in the Raman spectra of molecular vibrons (intramolecular vibration) that allow performing more definitive diagnostics. Indeed, solid molecular nitrogen phases show the presence of two major vibron modes ( and ) [53] corresponding to different site symmetries which exist in these structures; these result from slightly different intramolecular bond lengths and intermolecular arrangement and, therefore, in different vibrational frequencies. Naturally, this distinction vanishes in the fluid phase, which results in a single vibron mode (close in frequency to ) and is accompanied by an abrupt small frequency change of . A discontinuity in the vibron frequency is a typical phenomenon associated with phase transitions in molecular materials; it usually stems from changes in intermolecular coupling (e.g., [54]). A broadening of the vibron mode can be used as an additional diagnostic of melting [28]. This additional broadening is because the molecules in the fluid phase experience time-dependent thermally-induced bong length variations related to instantaneous changes in the local environment. The ultimate test for the Raman melting diagnostics should be provided by other experimental techniques such as, X-ray diffraction measurements of the diffuse scattering ring [21, 55].

Raman spectroscopy of an ensemble of vibrational modes in the hydrogen isotopes (H2 and D2) has been extensively used for the past decades to probe the phase diagram of this system at very high pressures. The experimentally accessed portion of the hydrogen phase diagram (<310 GPa at 100 K [16]) reveals three major solid molecular phases: orientationally disordered phase I and two partially or completely ordered high-pressure low-temperature phases II and III [56] (Figure 2). The presence in the Raman spectra of a series of rotational transitions (rotons) is the manifestation of the fluid or solid phase with freely rotating molecules. The Raman rotational transitions are described by quantum statistics governed by —the total angular quantum momentum number [57]. The departures from this simple model increase gradually with compression. The transition in phase II (or broken-symmetry phase, BSP) occurs due to an increase in anisotropic intermolecular interaction, which results in a change in the rotational ground state, which is no longer spherical [58]. This quantum ordering transition is exhibited by drastic changes in the Raman spectra of rotons and the vibron-roton coupled modes (vibron sidebands) [59] and appearance of the restricted rotational motion modes (librons) [60]. The low-frequency Raman spectra further change in phase III, where a number of narrow libron/phonon modes supersede the roton mode remnants manifesting the transition to a classically orientationally ordered phase [61]. The behavior of the phonon mode (corresponding to the out-of-phase lateral translational intermolecular vibrations) is less informative for the phase transitions; the steady increase of this mode shows an increase in the intermolecular coupling with compression.

Although these abrupt changes in the rotational Raman spectra with pressure are quite revealing for diagnostics of the molecular rotational and orientational states, their observations are rather experimentally challenging. Thus, Raman (and also IR) spectroscopy of the molecular vibron (which is easier to observed) was used to map out the phase boundaries in the majority of studies (Figure 2). The vibron frequency shows an abrupt discontinuity at the transition to phases II and III, and also between phases II and III [61, 62]. The observations of additional solid phases of hydrogen suggested initially theoretically [64] and experimentally [62] remain illusive [63, 65, 66]. The transition to phase III at 300 K is expected to be observed above 180 GPa, but this regime remains unexplored. The value of the vibron frequency discontinuity varies from a few wavenumbers (<15 cm−1) to over 100 cm−1 for the I-II and II-III transitions, respectively (Figure 3). The increased value of the vibron discontinuity for the transition in phase III arises from a modification of inter- and intramolecular interactions related to orientational ordering and structural changes at the transitions [61]. Moreover, phase III is characterized by greatly increased IR activity, which evidences for the orientational ordering with asymmetric atomic site symmetries [6769].

Vibrational spectroscopy remains the key diagnostics of the state and structure of dense hydrogen to megabar (100 GPa) pressures. The structures of phases II and III still needed to be determined unambiguously since the available X-ray and neutron diffraction data [70, 71] are not sufficient for this purpose. Measurements of vibrational and optical properties of hydrogen have been extended to 315 GPa at 100 K [16], but the study at higher temperatures still remain technically very challenging. Indeed, there is no published report on hydrogen above 200 GPa at room temperature.

The important feature of the vibron behavior at high pressures is a steady decrease of both Raman (Figure 3) and IR active vibrons, which indicate the approach of molecular dissociation, as this phenomenon is related to a partial charge transfer due to an increased coupling of the molecular orbitals through the intermolecular bonds. It is interesting to note that the vibron softening, which was suggested as a “harbinger” of the molecular dissociation [72], is not a common phenomenon as I2 and Br2 do not essentially reveal it [33]. Instead, other modes have been shown to soften near the transition to a monatomic state [33]. Future studies of hydrogen (H2 and D2) will show if this low-frequency phonon softening is a common feature of the transition to monatomic state or other more diffusive phase transition mechanisms can be also realized.

Given a great success of Raman spectroscopy for mapping the phase diagram and unraveling the bond properties of hydrogen at high pressures and low temperatures, one might expect that Raman spectroscopy can also be effectively applied for studies of hydrogen at high pressures and temperatures approaching the fluid phase and the plasma transition [7375]. The hydrogen vibron has been shown to soften substantially at high temperatures [76], and the melting transition has been diagnosed from a small vibron discontinuity [77]. Recent Raman study of hydrogen in the laser heated DAC revealed a large negative vibron frequency discontinuity near the melting line [24] (Figure 4). This has been accompanied by a large broadening and decrease in intensity of the roton bands [24]. These results suggest the existence of a pressure-induced transformation in the fluid, which might be related to the presence of a temperature maximum in the melting line as a function of pressure [73, 77, 78]. Previous Raman studies in the laser-heated DAC [76] also suggested the existence of such fluid with modified intramolecular bonds; moreover, it has been proposed that pairing fluctuations of nuclei may play an important role in such fluid states [79]. Future studies in the high regime probing these states may require an improved diagnostic techniques and perhaps a different experimental strategy (e.g., pump-probe methodology). Indeed, classical Raman spectroscopy can only provide information about the time-averaged material properties, while the information about the dynamical properties appears to be very important in this case.

3.2. Chemical Reactivity of Simple Molecules under Extreme Conditions

Knowledge of the physical and chemical properties of simple molecules such as water and methane under extreme conditions of pressure and temperature is essential to accurately model processes occurring in the interiors of a range of planets including Earth, Uranus, and Neptune [80]. Similar conditions are also generated by shocks and high explosive detonations [81]. It has been suggested that ice VII might also exist in cold subducting slabs in the Earth’s mantle [82]. The high magnetic fields of Uranus and Neptune measured by the Voyager 2 spacecraft have been attributed to conductive hot ices [83]. The speciation of carbon in the Earth’s interior is critical for understanding the origin of petroleum [84]. Formation and stability of methane under conditions corresponding to the Earth’s upper mantle have been demonstrated in recent years [8486] but the formation of heavier alkanes still remains controversial.

Raman spectroscopy in resistively and laser-heated DAC has been proven to be very informative for study of chemical transformations of simple molecules under extreme conditions. It is especially true for the molecules, which contain hydrogen as another popular technique—X-ray diffraction—is commonly not informative about the crystallographic positions hydrogen takes because of very low X-ray form factor of hydrogen.

The strength of Raman spectroscopy for water is that it is capable to probe the -dependent intra- and intermolecular bonding (e.g., hydrogen bonds); moreover, phase transformations and related to them orientational order-disorder phenomena can be readily recorded [29, 87]. Here we will concentrate on the behavior of water in the high-pressure limit of dissociation and ionization. Ice VII is a dense orientationally disordered cubic phase which is stable in a wide stability field (2–55 GPa). At higher pressures protons first become dynamically disordered along the line connecting the oxygen atoms (ice VII’, [88]), and then take the central positions between oxygens (symmetric ice X, [89]). Raman spectroscopy (along with IR spectroscopy, see [9093]) was very useful in unraveling this behavior, which was identified through the observations of the softening of the O-H modes associated with the strengthening of the hydrogen bonds, broadening of all vibrational modes and their concentration at low frequencies, and finally the appearance of the characteristic vibrations of the ionic ice X [94].

Raman spectra of hot ice VII show drastic changes upon heating in the region of the O-H stretch [19]. The frequencies of components of the O-H band increase with temperature and the spectral weight moves to higher frequencies (Figure 5) indicating that the hydrogen bond in ice VII weakens as the melting transition is approached and molecules acquire larger translational disorder. To determine the melting of ice VII, the behavior of the translational (phonon) mode has been monitored. This band broadens significantly and changes shape (additional intensity appears at lower frequencies as the result of a breakdown of the wave-vector conservation rule) when melting occurs, while in the solid phase the corresponding degree of temperature-induced broadening is very moderate. Concomitantly, changes also occur in the O-H band: a broad doublet is observed in the fluid that has a different intensity distribution compared to that in ice VII (see also [95]). The phonon mode of ice VII broadens substantially at 53 GPa and 300 K and upon heating at lower pressures (see also [94]) due to transformation to a dynamically disordered ice VII’; these changes allow tracing the corresponding transformation line.

The phase diagram of water determined by Raman spectroscopy under high conditions show the presence of a triple point between fluid water, ice VII, and ice VII’ (see also [96, 97]). Accordingly, the change in the slope of the melting line at this point suggests major differences between thermodynamic properties of ice VII and ice VII’. Therefore, it has been speculated that ice VII’ at high temperatures is superionic—the solid state with highly mobile protons and ordered (in a bcc lattice) oxygen ions [98]. Fluid water in this compression regime is expected to be highly ionized given the progressively increased difference in density between solid and fluid state and very weakly pressure-dependent melting curve near the triple point [55]. It has been inferred that water molecules are essentially dynamically dissociated under these conditions as the molecular lifetime becomes comparable to the vibrational period [19]. The Raman spectroscopic data summarized here are firmly supported by the results of the first principle theoretical calculations [19, 99]. However, the differences in melting temperatures determined in different experiments [19, 96, 97] and between theoretical calculations and some experiments [100] call for continuing studies.

Raman spectroscopy is very sensitive to monitor the chemical reactivity even in the case when minuscule amount of materials has experienced the transformation. Raman spectroscopic study of the laser-heated methane under upper-mantle conditions revealed the formation of saturated hydrocarbons containing 2–4 carbons (ethane, propane, and butane) and molecular hydrogen and graphite [101]. Conversely, exposure of ethane to similar conditions results in the production of methane (in addition to heavier hydrocarbons), suggesting that the synthesis of saturated hydrocarbons is reversible. These results support the suggestion that hydrocarbons heavier than methane can be produced by abiogenic processes in the upper mantle [101] (see also [102]). Prior to this work, ethane synthesis has been demonstrated in laser-heated DAC experiments on methane above 1200 K at pressures higher than 10 GPa [103].

The Raman signatures of heavier hydrocarbons are weak but extremely robust as the spectra in the spectral range corresponding to the skeleton C-C vibrations are very individual for hydrocarbons with different carbon compositions, intramolecular bonding, and molecular geometry (cyclic or chain-like). Moreover, this identification is confirmed by observations of the C-H deformation modes, which differ for the saturated hydrocarbons of different length.

3.3. Metals, Magnetic Materials, and High-Temperature Superconductors (HTS)

Raman studies of metals are a challenging task because only a very thin surface layer of the sample (skin depth is smaller than 1000 Ǻ) is typically probed by the exciting laser radiation. The author refers the reader to our previous review papers on the subject [5, 6]. Although the experimental capabilities remained essentially unchanged since the time of publishing of these review papers, the field has progressed substantially as new research has been performed on samples that were well characterized by other techniques. Here the author reviews recent works on metals (Co and Os), high-temperature superconductors (Bi2Sr2Can − 1CunO2n + 4 + x ), and a parent compound for the colossal magnetoresistance materials (LaMnO3).

Application of Raman spectroscopy to measurements of the shear elastic modulus in hcp metals has been discussed in great details on examples of Fe and Re [6, 10, 11]. Unlike -Fe (which is nonmagnetic), Co represents an interesting case of ferromagnetic metals, which is expected to transform to a nonmagnetic fcc phase (β-Co) under very high pressures (>105 GPa) [104], but the behavior of the magnetic moments as a function of pressure remained unknown until quite recently because use of the direct experimental technique (X-ray magnetic circular dichroism, XMCD) is challenging under high pressures [105]. Recent XMCP measurements show that Co remains magnetic in the entire pressure domain of stability of ε-phase, and the magnetic moments show a steady decrease with pressure [106]. Impulsive stimulated light scattering and Raman spectroscopy measurements revealed a departure from a common linear behavior of the elastic constants as a function of density [107], which indicate a gradual loss of magnetic moments in approaching the transition to β-Co (fcc). This softening of the elastic moduli with pressure has been also confirmed by the results of the inelastic X-ray scattering measurements [108]. The behavior of the vibrational frequency and the elastic constants are essentially different for the isostructural Co and Fe [107] (Figure 6), which makes the use of Co as the analog of Fe for single-crystal studies under high pressures problematic.

High-temperature superconductors (HTS) and materials with large electronic correlations reveal sizable electronic Raman scattering (ERS) cross-sections [109]. This is unlike the situation in metals, which normally show a very weak ERS because the probed momentum space q~1/δ  ( is the penetration depth, which is usually of the order of 10–100 nm for visible wavelengths) is very small and because of the screening of the charge fluctuations. Nevertheless, transition metals have been found to exhibit relatively strong ERS due to the presence of anisotropic Fermi-surface (FS). This allows a number of interband transitions to be visible by Raman spectroscopy in the wide frequency range ( is the electron Fermi velocity) under an assumption of ( is the frequency-dependent damping of the electronic state), that is in a clean limit [110]. Thus, Raman spectroscopy can be used as a tool to study the FS topography. For this, ERS should be studied for different values and directions. These can be controlled by changing the excitation wavelength and by choosing the scattering geometry. In addition to ERS, the phonon mode behavior can be studied. The phonon in Os demonstrated an anomalous -dependent softening, broadening, and asymmetry at low temperatures [111]. This finding suggests the occurrence of the Landau damping threshold [112]. The effect occurs in the momentum range corresponding to a crossing of the electron and phonon excitation energies.

The application of pressure may modify both the electronic band structure near the Fermi level and internal scattering processes, which would affect the Raman spectra. An anomalous increase in the electronic light-scattering cross-section was found in Os the pressure range of 20–30 GPa with green and blue excitation wavelengths [113]. At these conditions, an appearance of well-defined electronic peaks at 580 cm−1 for the wave-vector direction   [10 0] has been observed. These observations provide evidence for the increase in the damping of the electronic states for large wave vectors q, an essential part of which is probed by the high-energy excitation. Moreover, the anomalies of the optical phonon self-energies were also found in the pressure range of 20–30 GPa. The first-principle band-structure calculations suggest changes in the FS topology that occur due to the phonon displacements (Figure 7). Thus, one can speculate that these phenomena are responsible for the anomalies observed by Raman spectroscopy [113] and in the lattice parameters [114] above 20 GPa.

Raman spectroscopy is very sensitive to the local atomic structure even in the case of a transient local order. This has been exploited in a Raman study of LaMnO3—the parent compound for many doped manganites which exhibit colossal magnetoresistance (CMR). High-pressure, low-temperature Raman measurements performed on LaMnO3 up to 34 GPa provide the first experimental evidence for the persistence of the Jahn-Teller distortion over the entire stability range of the insulating phase [115]. This result demonstrates that LaMnO3 is not a classical Mott insulator as has been suggested earlier [116]. Domains of distorted and undistorted octahedra are present up to 32 GPa, and a weak signal from distorted octahedra is even observed in the metallic state at 34 GPa at 10 K. The insulator-metal transition begins when the number of symmetric octahedral domains increases beyond a critical threshold. The Raman spectra of a peak corresponding to undistorted octahedra show the asymmetry above 32 GPa, which may be associated with a Fano resonance occurring at the onset of the metallic phase, a phenomenon which has been previously observed in several manganite compounds [117, 118].

In HTS, Raman spectroscopy is an excellent tool to probe the electronic properties, including the electron-phonon coupling. Moreover, as a contactless probe separable from its surroundings, Raman spectroscopy circumvents the difficulties that prevent conventional methods from investigating a quantum critical point under pressure. Pressure, being a continuous, reversible, and laboratory-controlled physical variable, can tune properties of a single sample thereby studying the phenomena of interest in details. For example, pressure has been applied to underdoped insulating parent compound of the cuprates Bi1.98Sr2.06Y0.68Cu2O8 [119] to determine the generic phase diagram of HTS by tuning the carrier concentration. Raman measurements of the electronic background and collective modes have been used to extract the information about the electronic scattering, short-range magnetic correlation, and electron-phonon coupling. The results show the presence of a single anomaly in electronic, magnetic, and structural properties near 21 GPa. In conventional theory, magnetism, carrier transport, and superconductivity should all show separate transitions. These results establish the presence in copper oxides HTS of a quantum critical point inside the superconducting dome in close similarity to other unconventional superconductors in strongly correlated materials.

In another high-pressure study, pressure has been applied to an optimally doped HTS Bi2Sr2Ca2Cu3O10 + δ with three copper-oxygen (Cu-O) conducting layers in a unit cell. The combined superconducting critical temperature ( ) (using magnetic susceptibility) and Raman spectroscopic studies (Figure 8) under hydrostatic pressure up to 37 GPa reveal an unpredicted superconductivity enhancement above 24 GPa [120]. This observation suggests the presence of a crossover from the competing order to superconductivity in the inner CuO2 plane. The latter increase occurs as a result of competition between pairing and phase ordering in different CuO2 planes (e.g., [121]). The observations have important implications for engineering HTS with much higher ’s at ambient conditions.

3.4. Pressure Metrology

Raman pressure gauges have attracted a great interest recently [123126]. These commonly include diamond (usually 13C) and cubic boron nitride (cBN). These materials are very stable structurally and chemically under extreme conditions. They possess a sizable Raman signal when used as the pressure gauges in the DAC pressure chamber under extreme conditions and this signal can be easily separated spectrally and spatially from that of the stressed Raman anvils. These two factors give the Raman pressure sensors important advantages over more widely used fluorescence gauges [127, 128]. Moreover, Raman measurements of the stressed diamonds have been proposed for pressure measurements (e.g., [123]), thus eliminating the need of having a pressure gauge inside the pressure chamber.

Calibrations of these Raman sensors present a difficult problem under the conditions of very high pressures and extreme temperatures. The common strategy is to perform concomitant measurements on a number of pressure gauges in as close as possible hydrostatic pressure conditions. Moreover, the materials themselves can be used as primary calibrants. An accurate pressure scale can be constructed based on determination of pressure from simultaneous density and sound velocity measurements (e.g., [129]) under static conditions using the so-called redundant scheme. This method requires only a relatively small correction from an adiabatic to an isothermal path at room or moderately elevated temperature, so from that point of view it is expected to be more accurate than shock-wave based calibrations [130]. In situ high-temperature concomitant Brillouin spectroscopy and X-ray diffraction measurements can be performed on the crystallographically oriented singe-crystal specimen. Raman measurements can be performed concomitantly. These experiments are challenging and time consuming; moreover, they require specialized facilities such as, the Brillouin-X-ray bending magnet beamline at the sector 13 at the Advanced Photon Source (APS) at ANL. Such experiments are currently in progress [131].

For the purpose of this paper the author presents here the review of most relevant Raman works on diamond and cBN. The experiments at room temperatures under quasihydrostatic conditions are currently limited to approximately 150 GPa [132, 133]. The frequency data show a small curvature in coordinates, and they are essentially linear as a function of density, ρ [133, 134]. Thus, the mode Gruneisen parameter is effectively pressure independent. Even at the highest pressure probed, the samples remain in a low-compression regime ( ) making expansions in power series very good approximations to represent both the equation of state (EOS) and Raman shift with pressure. Thus, combined Raman spectroscopy-density measurements constrain the high-pressure scale at room temperature. The effect of cooling below 300 K is negligible on both the EOS and the Raman frequency, so this conclusion holds at low temperatures.

Pressure calibration at high temperature is of special interest, as many of the experiments related to geosciences, planetary sciences and other disciplines require reliable pressure measurements at high temperatures. The synchrotron X-ray diffraction and Raman spectroscopy measurements have been performed in cBN to various pressures depending on temperature using a combination of laser, internal, and external resistive heating in the DAC [122, 132, 135, 136]. These data have been recently used to calibrate the fluorescent pressure sensors up to 120 GPa and 700 K [137]. Laser-heating experiments in cBN to 1750 K at a maximum pressure of 40 GPa, and up to 2300 K at 26 GPa to 40 GPa (Figure 9) show that the phonon frequencies can be well modeled by assuming a pressure-dependent quasiharmonic temperature contribution combined with an almost pressure-independent intrinsic contribution [122].

At very high pressures (>150 GPa) measurement of the Raman shift corresponding to the most stressed part of the anvil [138] is arguably the most viable technique for pressure measurements. Although the results have been shown to be dependent on the experimental geometry (culet dimensions [139], and loading axis direction [140]), other techniques suffer from a necessity to have a specialized equipment at the time of experiment (e.g., a tightly focused X-ray synchrotron beam) and from even larger calibration uncertainties. Thus, the Raman technique remains a very practical alternative.

4. Conclusions and Outlook

Raman spectroscopy has been established as one of the most used and powerful technique for high-pressure research in the DAC. The recent developments have improved the sensitivity, the spatial resolution, and, as the result, made it possible to use it as a rapid diagnostic tool including the use for in situ laser heating measurements and as portable systems, such as online systems at X-ray synchrotron beamlines. There are prospects for further improvements of the spatial and temporal resolution of the Raman systems applied for high-pressure research. Moreover, very high-throughput low-frequency (<10 cm−1) Raman systems will soon become routinely available.

These technical improvements have enabled multidisciplinary studies, which include material science, physics and chemistry of planetary interior, search for alternative energy sources and carriers, detonation physics and chemistry, and many others. The improved spatial and temporally resolutions will enable study of the chemical reactivity in heterogeneous samples (including mineral interfaces) and other fast processes. This progress is tightly connected to new capabilities of the surface enhanced Raman spectroscopy [141], which remain to be developed. Moreover, the capabilities of time-resolved and pump-probe techniques remain to be further exploited; the use of coherent anti-Stokes Raman spectroscopy (CARS) with the dynamic high-pressure techniques [142, 143] will broaden high-pressure and geoplanetary science applications.

Future prospective scientific directions will call for new developments of the Raman techniques that permit direct probe of these properties in situ statically or dynamically using advanced laser sources and detecting techniques. Those include ultrafast lasers and time-resolved detectors with improved spatial and temporally resolution (e.g., streak-cameras). Of particular interest is to develop new capabilities to probe material properties in extremely small quantity and extreme environments that are beyond current reach but could potentially provide new insights into the materials synthesis under nonequilibrium conditions and into the planetary history and structure.

Acknowledgments

The author acknowledges financial support from NSF Earth Sciences (EAR-0842345 and EAR-1015239), Carnegie/DOE Alliance Center (CDAC), Energy Frontier Research in Extreme Environments (EFree) Center, NASA Astrobiology Institute (NAI), Army Research Office, Carnegie Canada, and Carnegie Institution of Washington. The author thanks R. J. Hemley, V. V. Struzhkin, H.-K. Mao, M. Somayazulu, E. Gregoryanz, X.-J. Chen, M. Baldini, N. Subramanian, V. Prakapenka, Yu. Ponosov, J. C. Crowhurst, B. Mysen, and D. Foustoukos for the valuable comments on the paper.