A CO₂ laser heating system for in situ high pressure-temperature experiments at HPCAT

Experiments in the laser-heated diamond anvil cell (LH-DAC) are commonplace within the extreme conditions community and allow access to temperatures up to thousands of degrees whilst at static high pressures up to millions of atmospheres.^1,2 The demand for laser heating at static high pressures has been recognized since the 1970s, with early efforts focusing on the CO₂ laser.³ The later-developed high-power solid-state near-infrared lasers became popular for extreme conditions experiments and saw the development of powerful instruments by combining the LH-DAC with synchrotron x-ray techniques.^2,4–13 Although less frequently utilized, CO₂ laser heating has remained present in extreme conditions experiments,¹⁴ particularly in mineral physics,^15–17 and combined with Brillouin scattering,^18,19 but there are comparatively few examples of the technique performed in situ with x-ray diagnostics at a synchrotron.^20–26 Past examples of CO₂ laser heating systems at high-pressure diffraction beamlines date to the 1990s at the Cornell High Energy Synchrotron Source (Ithaca, NY, USA),²⁷ Laboratoire pour l’Utilisation du Rayonnement Electromagnétique (Orsay, France),²⁸ and the ID-30 beamline of the European Synchrotron Radiation Facility (ESRF; Grenoble, France),²⁹ and modern examples can be found at the SPring-8 synchrotron (Hy$o¯$go, Japan),³⁰ and the ID-27 beamline of the ESRF.⁹ 

Solid-state lasers with wavelengths around 1 μm are commonly used for heating metallic samples within the DAC. The near-infrared radiation is absorbed by free electrons in the metal via a reverse-Bremsstrahlung mechanism, oscillating within the electric field of the incoming radiation and transferring heat to the lattice through collisions.³¹ In some cases, e.g., Fe-bearing dolomite³² and (Fe, Mg)SiO₃ perovskite,³³ the presence of small amounts of Fe can provide sufficient absorption at 1 μm to facilitate direct heating compared with their Fe-free counterparts. In most wide bandgap materials, however, lasers of these frequencies do not provide effective heating. One common practice is to mix sample materials with some metallic coupling material, thereby providing indirect heating to the non-metallic sample, but this method comes with its own issues: The indirect nature of the heating means that information regarding reaction kinetics may only be inferred, the introduction of a metal into the sample material provides the risk of unwanted chemistry, and the use of a relatively high-Z coupling material drastically limits the quality of x-ray diffraction data accessible for lighter materials.³⁵ Additionally, large temperature gradients which facilitate chemical diffusion and segregation within laser-heated materials have been observed when heating is facilitated only by some inhomogeneous impurities—as in Fe-bearing minerals—and when metallic absorbers are employed.³⁴ 

The use of a CO₂ laser allows non-metallic sample materials to be heated directly. 10.6 μm radiation equates to a photon energy of 943 cm⁻¹, which is of the same order as lattice phonons in covalent crystals, and is absorbed by existing lattice vibrations. This absorption mechanism also provides the advantage of more uniform heating—the materials concerned are generally transparent to 10.6 μm, meaning that heating is provided via this process throughout the entire thickness of the sample, unlike in metallic systems where the laser radiation is absorbed primarily at the surface. The long wavelength of CO₂ laser light results in a significantly larger focused spot than its near-infrared counterparts, providing a more uniform heating across the sample laterally with less pronounced temperature gradients which can influence data collection—focused laser spots are typically an order of magnitude greater than the size of the x-ray probe. No requirement for a metallic coupler allows potentially detrimental extraneous materials to be omitted from the sample arrangement.

We present here the recently developed instrument for direct heating of wide bandgap insulating materials via a CO₂ laser inside a diamond anvil cell (DAC) at a dedicated synchrotron x-ray diffraction beamline (16-ID-B, HPCAT, Sector 16, Advanced Photon Source, Argonne National Lab, Illinois, USA). The system is designed for portability between experiment hutches, aligning with the wide variety of x-ray diagnostics available at HPCAT.

An overview of the optical schematic is displayed in Fig. 1. It comprises four separate components, indicated by the wavelength legend: laser delivery, x-ray diffraction, optical pyrometry and imaging, and mid-infrared microscopy. With these, a complete alignment of the system can be achieved by first bringing the DAC to the focus of the x-ray beam, visualizing the DAC interior with the optical pyrometry system and mid-infrared microscope, and finally aligning the CO₂ laser to the sample. The x-ray diffraction capabilities of the 16-ID-B diffraction beamline at HPCAT (Sector 16, Advanced Photon Source, Argonne IL, USA) have been discussed in depth elsewhere in Ref. 36. 

The laser takes the path shown by red traces in Fig. 1. A Synrad Evolution125 CO₂ is installed on the “general purpose” optical bench of the ID-B diffraction hutch at HPCAT (Advanced Photon Source, Argonne National Lab, Illinois, USA). The Synrad UC-2000 laser controller may be controlled via Experimental Physics and Industrial Control System (EPICS) for remote operation. The optical coupler of the Synrad laser features a square aperture, and the outgoing beam is subject to diffraction from it. Beyond the Fraunhofer diffraction limit, the beam tends toward a circular sinc²(x) function, with M² < 1.2. Refracting optics are not placed in the beam path until beyond this limit—in this case 60 cm. At its output, the beam has a waist of 4.4 mm and a full-angle divergence of 3.2 Mrad.

The focusing beam path is introduced off-axis by 30°, allowing for on-axis visualization of the DAC interior and collection of thermal emission. The 30° angle-of-incidence is selected based on typical openings of modern diamond anvil seats dedicated to x-ray techniques. Additionally, power density supplied to the sample may be maximized in the off-axis geometry by manipulating polarization of the laser light to minimize reflections from the table of the diamond. For an angle of 30°, and extrapolating the refractive index of diamond to 2.378 at 10.6 μm from Ref. 37, the Fresnel reflection from the table of the diamond may be reduced from 16.79% for randomly polarized to 12.69% for purely p-polarized light—equating to >5 W of additional laser power delivered to the sample at the maximum output.

Controlling the position of the laser spot within the DAC is commonly achieved by two methods: translating the focusing lens (ZF) or placing the steering mirror in its converging path. Using the former, one must consider the waist of the beam and the required motion of the lens. At 30° angle-of-incidence, moving the focused spot by a distance of ±250 μm relative to the center of a 1.7 mm-thick diamond anvil (typical of a Boehler-Almax design anvil³⁸) requires translation of the lens by ±3.5 mm—owing to the fact that the beam is refracted from 30° to 12.14° across the air-diamond interface. This bound on the range of motion significantly limits the maximum waist of the beam at the focusing optic. For example, for a Thorlabs LA660-G ZnSe plano-convex lens (chosen for ZF), the minimum spot size according to the diffraction limit is 52.6 μm when the entire clear aperture (20.32 mm) is filled, compared with 76.6 μm when the lens is only partially filled to accommodate this motion.

We therefore opt for a fixed lens assembly with a steering mirror to translate the focused spot. A 4.0× beam expander (BE; ULO Optics C-BE4.0) expands and collimates the primary laser beam to fill the clear aperture of the focusing lens (ZF; f = 75 mm)—reducing the minimum focused spot size by almost 50%, from 98.5 μm to 52.6 μm. The fine motor control on the steering mirror (SM) allows movement of the laser spot within the DAC with high precision. Meanwhile, ZF is motorized along the laser axis to allow for control over the power density delivered to the sample by softening the focus of the laser spot.

Temperature measurements are performed using the spectral analysis of thermal emission in the visible region (yellow paths in Fig. 1).

The imaging and pyrometry system is fronted by a polished amorphous carbon mirror with a thin Ag coating (AC), allowing imaging to be performed on-axis with minimal effects on the x-ray beam. A MgF₂ window—strongly absorbing at 10.6 μm³⁹—is placed before the collecting lens (CL; f = 75 mm) in order to protect the optic from stray laser light. The collected light is refocused with a magnification of 6.67× (RL; f = 500 mm) onto a mirror pinhole (MP), which acts as a spatial filter for optical pyrometry measurements:^6,9,36 the thermal emission which passes through the 50 μm pinhole may be spectrally analyzed by a fiber (OF)-coupled Ocean Optics HR2300 + ES spectrometer (calibrated to the thermal emission of a NIST-calibrated tungsten filament), while the remaining light is reflected. 6.67× magnification and a 50 μm pinhole equate to a sampling area diameter of 7.5 μm over which temperature analyses are performed, comparable to the size of the x-ray beam.

The reflected light is re-imaged by a Navitar Zoom 6000 zoom lens (ZL; working distance = 100 mm) via a 50:50 beam-splitting cube (BS; width = 10 mm) onto a Watec 137LH CCD camera. A partially reflecting pellicle on a pneumatic actuator (PP) can provide illumination from an LED to visualize the interior of the DAC and can be removed during pyrometry measurements to avoid thin-film interference patterns in the collected spectra.

The optical pyrometry and visualization system is compact (Fig. 2) and lightweight (6.5 kg) and can be easily transported between the various experimental hutches at the HPCAT beamline.

Perhaps the most limiting factor when working with CO₂ laser radiation is the lack of visualization. While the quantum efficiency of CCD or CMOS cameras will allow 1 μm radiation from a solid state laser to be imaged, wavelengths in the mid-IR have remained difficult to directly observe. Common practices to align the focused laser spot in CO₂ laser-heating arrangements include having the laser coupled with the sample or some other material inside the sample chamber to achieve thermal emission and aligning the “hot spot,” though this carries the inconvenience of requiring temperatures >1000 °C in order to align the laser. Other practices include the use of a “dummy” DAC, whereby the CO₂ laser is aligned via transmission to a drilled gasket which behaves as a pinhole. However, with the significant refraction by the diamond anvils, slight differences in DAC geometry reduce the accuracy of this indirect alignment method.

We employ a mid-infrared microscope to directly visualize the interior of the diamond anvil cell in the mid-IR region. The microscope comprises a FLIR thermal imaging module—a two-dimensional array of vanadium oxide bolometer “pixels,” sensitive to 8-14 μm radiation. At present, we have worked with the FLIR Lepton and Boson imaging modules. The Boson module offers greater resolution (12 μm pixel pitch vs 17 μm), has a larger sensor (640 px vs 80 px), and features digital zoom.

The imaging module has its stock wide-angle lens removed in order to expose the detector, and a pair of ZnSe plano-convex lenses (Fig. 1 inset) is used to form an image with 2× magnification. A broadband mid-infrared source fitted with an Al parabolic reflector (IR; Hawkeye Technologies IR-12) is used to coaxially illuminate the sample environment via Fresnel reflection from a ZnSe window (ZR). The mid-infrared microscope is fixed on the same optical breadboard as the optical pyrometry and visualization system, and manually placing an Au mirror before the amorphous carbon mirror (inset of Fig. 2) allows the microscope to see inside the DAC. Figure 3(a) shows an image of a 180 μm sample chamber drilled through a gasket indented using diamond with 300 μm culets formed on the FLIR Lepton module. The artifact on the left of images is due to previous damage caused by intense CO₂ laser light—this is often reversible over time but can cause permanent damage. Notably, the presence of CO₂ laser light in the image causes features due to the broadband IR source (IR) to be washed out. At present, the minimum lasing power of the Synrad Evolution 125 laser far exceeds the light intensity from IR, and alignment is performed first by having the mid-infrared imaging module display the gasket and sample chamber of the DAC [Fig. 3(a)] and then aligning the CO₂ laser spot to its location [Fig. 3(b)]. In Fig. 4, the mid-infrared microscope with Boson module images the entire table of the diamond anvil with the CO₂ laser at extremely low power. The reflection from the gasket indentation can be seen in the center of the image when the laser is incident on the sample area, as well as the Fresnel reflection from the diamond anvil table (red dashed circle). By digital or optical zooming, a fine alignment of the CO₂ laser spot to the exact location of the sample can be achieved.

We find it useful to keep an additional FLIR Lepton module with its stock 50° field-of-view lens inside the experimental hutch on a mobile basis. This device may be moved around the hutch to track the position of the CO₂ laser beam along its optical path, to observe the back of the DAC for transmission of laser light as a secondary means of confirming alignment, and to properly locate stray reflections in the hutch to ensure safe encapsulation.

To demonstrate the capabilities of the instrument, we have selected zirconia (ZrO₂) as a model system. We performed CO₂ laser heating on ZrO₂ in a DAC of custom design, fitted with conical-cut diamonds,³⁸ which facilitate the 30° angle for the large incoming CO₂ laser beam.⁶ A gasket was prepared by indenting 200 μm-thick Re foil to a thickness of 30 μm and creating a 180 μm sample chamber by laser micro-machining.⁴⁰ ZrO₂ (99% metal basis excluding Hf, Alfa Aesar) was surrounded by dry KBr (Sigma-Aldrich) to serve as a pressure-transmitter and thermally insulate the sample material from the diamonds—KBr was chosen for its high melting curve, comparable to ZrO₂.^41,42 Two-dimensional x-ray diffraction patterns were integrated using the Dioptas software package.⁴³ 

Room-temperature compression of ZrO₂ has the material transform from its ambient baddeleyite (P2₁/c) phase into a distorted fluorite (Pbca, “ortho-I”) structure above ∼6 GPa,⁴⁴ followed by a transformation into a cotunnite-type (Pnma, “ortho-II”) structure above 12.5 GPa which persists until beyond 100 GPa and up to 2500 K in near-IR laser heating experiments.⁴⁵ 

Pressure-induced first-order phase transitions may be hindered by kinetic barriers and thus highly dependent on energetic pathways and can be affected by compression rates⁴⁶ and hydrostaticity.⁴⁷ Laser annealing has previously been shown to greatly reduce pressure gradients and improve hydrostaticity inside DAC sample chambers.⁴⁸ Through CO₂ laser annealing, we have recently demonstrated the capability of overcoming kinetic barriers which have otherwise obscured the existence of a high-pressure phase in CaCO₃.⁴⁹ 

In Fig. 5, annealing with low laser power density—such that the temperature is below the sensitivity of visible optical pyrometry as a reliable temperature sensor—vastly improves the quality of data. A room-temperature compression of ZrO₂ in a KBr medium sees the Bragg peaks broaden significantly as anisotropic stresses are applied to the powder sample and it nears a first-order phase transition. The ambient temperature pattern in Fig. 5 is thus difficult to subject to any rigorous structural refinement. By annealing with the CO₂ laser, the residual stresses within the sample are relaxed, and this is evidenced by the improved sharpness of diffraction features, especially at higher angles, and that we are able to perform a Rietveld refinement on the two coexisting phases of ZrO₂ at this pressure.

Below the ortho-II phase transition, zirconia has a tetragonal (P4₂/nmc) phase which exists at temperatures above around 1000 K.⁴² At increasingly high temperatures, ZrO₂ has been presumed to adopt a cubic fluorite (⁠$Fm3¯m$⁠) structure, but recent experiments in a laser-heated DAC did not detect any onset of this phase up to 3000 K at 12.5 GPa,⁴⁵ leading those authors to speculate a ZrO₂ phase diagram with no high temperature cubic phase.

In Fig. 6, ZrO₂ is heated using the CO₂ laser to 2159 ± 13 K with x-ray diffraction measured in situ on the heated sample. Note that a full structural Rietveld fit could be performed on the sample at these pressure-temperature conditions. The sample was heated to a maximum ZrO₂ temperature of 2860 ± 6 K and—as in Ref. 45—we saw no evidence of the fluorite structure, which had been postulated to exist at these conditions.

We have developed an instrument for direct heating of wide bandgap insulating materials inside a diamond anvil cell, further broadening the range of techniques available at the HPCAT synchrotron beamline. The instrument is designed around portability between experiment hutches at the beamline, with hopes for combined CO₂ laser heating and x-ray absorption spectroscopy in the near future—a powerful tool for observing melting phenomena at high pressures which has thus far only been achieved in metallic systems with 1 μm laser heating.⁶ The implementation of mid-infrared microscopy allows for more controlled alignment of the CO₂ laser by directly visualizing the sample environment and the low-power laser spot. In future iterations, it is conceivable that the same mid-infrared sensor currently used for qualitative purposes (imaging and alignment) can also be used for quantitative analyses (temperature readout and pyrometry in the room-temperature regime). The absorption mechanism for 10.6 μm radiation (Sec. I) means that absorption lengths are typically tens of microns, as opposed to the near-surface absorption in 1 μm laser heating systems, and uniform heating can be achieved with only single-sided laser illumination. This allows space between the DAC and imaging plate (Fig. 1) for installation of a multi-channel collimator. Future iterations of the instrument will thus benefit from significantly reduced background from the Compton scattering of the diamond anvils, allowing measurements to be made on CO₂ laser heated samples with a low atomic number, even within the fluid state.⁵⁰ 

This research was sponsored in part by the National Nuclear Security Administration under the Stewardship Science Academic Alliances program through DOE Cooperative Agreement No. DE-NA0001982 and in part by the U.S. DOE under Grant No. DE-FG02-99ER45775 to GS. This work was performed at HPCAT (Sector 16), Advanced Photon Source (APS), Argonne National Laboratory. HPCAT operations are supported by DOE-NNSA under Award No. DE-NA0001974, with partial instrumentation funding by NSF. APS is supported by DOE-BES, under Contract No. DE-AC02-06CH11357.