Three Decades Later, Metallic Oxygen Finally Yields Its Raman Signature

For more than three decades, the vibrational properties of metallic oxygen resisted detection because of weak Raman scattering, high reflectivity and the demanding diamond anvil cell environment.

© Hübner Photonics

Raman spectroscopy is one of the most powerful techniques for investigating the structure of materials, yet some systems remain beyond its practical reach due to extremely weak scattering signals. Metallic oxygen is one such example. Although oxygen is known to become metallic at high pressures, its vibrational properties have remained largely inaccessible for more than three decades because of the combined challenges of weak Raman scattering, high reflectivity, and the demanding environment of diamond anvil cell experiments. This article shows that these challenges can be overcome through exceptional instrument performance. Using highly stable single-frequency laser sources at 532 nm, 660 nm, and 785 nm, the authors successfully recorded clear Raman spectra of metallic oxygen across a broad pressure range. All three wavelengths deliver high-fidelity spectra with minimal trade-offs, establishing new benchmarks for exploring matter at extreme conditions, proving that even the faintest material responses can now be effectively captured.

Background

Pressure is a powerful thermodynamic parameter for probing and tuning interactions in condensed matter. In molecular systems, compression reduces intermolecular distances, driving profound electronic and structural transformations. Under extreme pressure, insulating ambient dielectric gases can metallize and ultimately form close-packed atomic solids. These conditions are routinely achieved using diamond anvil cells, which generate pressures comparable to those at Earth's center while the diamond provides optical access across a broad spectral range, from X-rays to the far-infrared. It makes the DAC an ideal window to extremes, allowing for the in-situ interrogation of materials with a variety of spectroscopic probes.

One of the most celebrated goals of high-pressure physics is the creation of metallic hydrogen, predicted in 1935 and anticipated to exhibit remarkable properties including high-temperature superconductivity. While its synthesis remains elusive due to the vast pressures required, oxygen (a molecular analogue of hydrogen) metallizes at 96 GPa and has been reported to become superconducting below 0.6 K [1]. Despite its accessibility, the structure and vibrational properties of metallic oxygen (ζ-O₂) remain poorly understood [2].

Raman spectroscopy is uniquely suited to DAC experiments because it can probe the minute sample volumes of tens of cubic microns, at extreme pressures. It has played a central role in establishing phase diagrams and identifying crystal structures through lattice 

vibrations and offers a direct test for computational models predicting energetically favorable molecular configurations. However, high-pressure Raman measurements are intrinsically challenging due to optical aberrations and reflections from the diamond anvil. 

Metallic samples present an even greater challenge: strong electronic screening limits optical penetration depths and severely weakens the Raman response. As a result, the vibrational spectrum of metallic oxygen has resisted clear spectroscopic characterization for more than three decades, representing one of the most demanding problems in high-pressure spectroscopy.

Results and Discussion

Oxygen was cryogenically loaded as a liquid into a diamond anvil cell (DAC) at 100 K and 10 bar and sealed within a laser-machined rhenium gasket. High-purity Type IIas diamonds were used to minimize fluorescence and maximize optical clarity. With roughly 100 µm culets and a 30 µm sample chamber, pressures exceeding one million atmospheres were achieved, driving oxygen into its metallic phase. Raman measurements were performed in a confocal backscattering geometry using 532 nm, 660 nm, and 785 nm excitation focused to a ~2 µm spot using a high-numerical-aperture microscope objective. The scattered light was filtered by ultra-narrow notch gratings, then dispersed by a spectrometer and analyzed with a CCD detector.

Under compression, oxygen undergoes a sequence of molecular phase transitions as increasing orbital overlap closes the bandgap. A notable feature is the formation of O₈ molecular units, consisting of four O₂ molecules arranged in a harmonious quartet [3]. Above 96 GPa, oxygen develops a metallic luster and becomes visually indistinguishable from the surrounding rhenium gasket. Despite decades of study, the structure and vibrational properties of this metallic phase have remained unresolved due to extremely weak Raman signals and phase coexistence.

This paper reports on the first unambiguous vibrational spectrum of metallic oxygen obtained using multiple excitation wavelengths. The spectra reveal a rich phonon response between 50 and 800 cm⁻¹. After correcting for the Bose thermal population factor, which otherwise distorts line shapes below approximately 200 cm⁻¹, consistent peak positions are observed at 150 GPa for all three laser wavelengths. The Raman signal is extraordinarily weak; the mode near 400 cm⁻¹ is approximately 10¹² times weaker than the excitation line, requiring exposure times of up to 10 minutes at laser powers approaching 30 mW. 

Observation of anti-Stokes scattering confirms the first-order Raman origin of the signal, while wavelength-dependent intensity variations indicate resonance effects linked to the electronic structure.

Comparison with theoretical predictions enables a definitive structural assignment of metallic oxygen in this study. The results identify a crystal structure that preserves the distinctive O₈ molecular units found in the dense insulating precursor phase.

Lasers for low frequency Raman 

To obtain observable Raman signals in the fingerprint region under high-pressure conditions, the excitation source must meet several stringent requirements. First, excellent spectral purity is essential to enable detection of the weak low frequency Raman spectra close to the laser line. In addition, outstanding beam pointing stability, stable output power, and wavelength accuracy is critical, as the inherently weak Raman signal necessitates rather long acquisition times and a tightly focused beam within the diamond anvil cell, increasing the risk of misalignment.

The compact single-frequency Cobolt Samba 532 nm, Cobolt Flamenco 660 nm, and Cobolt Disco 785 nm  lasers are suited to meet these demands. These lasers combine spectral purity (>80 dB suppression) with a TEM₀₀ beam (M² < 1.1), enabling Raman measurements at low spectral shifts and efficient micro-focusing into the diamond anvil cell. Furthermore, these diode-pumped solid-state lasers deliver outstanding beam-pointing stability of 5 μrad/°C, power stability better than 2 percent, and wavelength drift below 1 pm over an 8-hour period. Together, these characteristics make Cobolt lasers suitable for high-pressure Raman spectroscopy applications.

Outlook

While the vibrational Raman signature of metallic oxygen is important, the significance of this work lies in overcoming a previously insurmountable technical limit. Measurements once considered impossible are now achievable, thanks to the availability of high-performance single frequency lasers, opening the door to studying a host of other systems with inherently weak Raman responses, including electronic Raman, superconducting peaks, and amorphous materials. By pushing past this barrier, we have added a powerful new tool to the researcher's arsenal for exploring matter at extreme conditions, proving that even the faintest material responses can now be effectively captured.