Implementation of extreme ultraviolet spectroscopy on a sheared-flow-stabilized Z pinch

Plasma spectroscopy is an important technique for the measurement of fundamental parameters in an experimental plasma device or fusion generator. The spectral signature of a radiating plasma contains information about the elements within the plasma, as well as the temperature and density. The plasma within many devices contains impurities due to plasma–material interactions, sputtering, and desorption processes. In fusion experiments, impurities in the fusion fuel enhance radiative power losses, increasing the required triple-product needed to reach net fusion gain.

The sheared-flow-stabilized (SFS) fusion Z-pinch experiment (FuZE)¹ investigates a low-cost approach to thermonuclear fusion at relatively moderate densities and temperatures. By embedding an axial flow with radially dependent velocity, m = 0 and m = 1 instabilities are suppressed in the Z pinch.^2,3 The FuZE device has been observed to produce sustained neutron production lasting 10 µs, stable quiescence period, radially dependent flow velocities, and >1 keV ion temperatures.^4–7 Neutron energy anisotropy studies found evidence of thermonuclear neutron production.⁸ Optical Thomson scattering experiments measured electron temperatures between 167 ± 16 and 700 ± 85 eV and electron densities of ≤5 × 10¹⁶ cm⁻³ in the FuZE plasma.⁹ 

To study the extreme ultraviolet (EUV) plasma radiation from the next-generation FuZE device, FuZE-Q,¹⁰ at Zap Energy, a time-gated vacuum-coupled spectroscopy diagnostic has been designed and fielded. This article describes the EUV spectrometer diagnostic in detail, along with the image and data processing techniques used for analysis. Finally, this article presents some initial parameters of the early FuZE-Q device in connection with spectroscopic results from the EUV diagnostic.

Spectroscopy is a non-perturbative technique to diagnose impurity levels, temperature, and density in plasma and fusion experiments. In magnetic-confinement fusion devices, the plasma-material interactions that occur on the divertor and chamber walls are often a source of high-Z impurities, such as tungsten, chromium, zirconium, copper, titanium, iron, nickel, and molybdenum.^11,12 Several lower-Z impurities, such as boron, lithium, carbon, nitrogen, oxygen, and silicon, are also very common.¹³ Noble gases, such as neon, argon, and krypton, can be used as tracers enabling additional spectroscopic analysis for fusion plasmas.^14–16 While some spectroscopy techniques diagnose radiation in the x-ray range, the majority of impurity line radiation is in the lower energy extreme ultraviolet (EUV/XUV) wavelengths, ∼1–50 nm. EUV spectroscopy techniques are important tools in many magnetic-confinement fusion experiments.^17–19 

A McPherson 310G 1 m diameter Rowland circle grazing incidence vacuum spectrometer²⁰ was fielded on an SFS Z-pinch experiment. The 310G spectrometer was equipped with a six-strip microchannel plate (MCP) and 600 g/mm toroidal platinum-coated diffraction grating. The MCP strips were imaged, by means of a high-voltage (HV) biased phosphor coated thin-film electrode, 3:1 fiber optic, and fast f/# lens setup, with a 16-bit CCD camera, as shown in Fig. 1.

Light from the entrance slit falls incident on the 600 g/mm toroidal diffraction grating with an incidence angle of 87° and is dispersed along the 1 m diameter Rowland circle. The platinum-coated 600 g/mm grating, blaze angle of 2°, is sensitive over a wide wavelength range. By the 2/3 to 3/2 wavelength rule, the grating has the highest relative efficiency between 6.76 and 15.21 nm wavelength. The grating has a blaze wavelength of 10.14 nm and a groove spacing (d) of 1666.67 nm. The grating width and height are 2.5 and 2 cm, respectively.

The diagnostic uses a 40 mm diameter, six-strip MCP; however, only three of the strips were utilized during experiments. The strips, made of gold, have a cesium-iodide coating to improve efficiency in the EUV regime. A 3:1 fiber optic reducer with a phosphor coated thin-film electrode, biased at 4250 V, faces the grounded MCP anode with a small gap. The 3:1 fiber optic reducer shrinks the image without significant loss of resolution. The MCP, along with the strips and phosphor, allows the EUV spectrum to be imaged in the visible wavelength range.

Visible light from the backside of the phosphor coating is collected by the fiber optic reducer. The output is then imaged with two f/1.2 lenses onto a CCD camera. The diagnostic uses a 1024 × 1024 pixels, 16-bit Marconi (E2V CM2-1) CCD.²¹ The pixel and chip size are 13 × 13 µm² and 13.3 × 13.3 mm², respectively. The CCD electron well depth is 100 000 with a dark current of 1635 electrons/pixel/s. The CCD is cooled via the Peltier effect to 0 °C to reduce noise. Dark current is typically reduced by a factor of two for every ∼7 °C below room temperature until the CCD reaches −10 °C. Cooling the CCD to temperatures of ∼0 °C is sufficient to reduce dark current to negligible levels for exposures shorter than $∼1$ ms. Even when the dark current is significant, for exposures much longer than 1 ms, the dark current noise is small.

Collecting data on the SFS FuZE-Q device presented many technical challenges for the EUV spectrometer diagnostic. The radiation of interest is significantly attenuated by any glass or composite vacuum window, and thus, the EUV spectrometer was vacuum-coupled to the experiment with a direct line-of-sight to the plasma. However, many challenges come with a vacuum-coupled design. The following paragraphs will describe these technical challenges and the steps taken to address them.

The rapid pressure changes in the SFS FuZE-Q experiment during a “shot” present a challenge for the EUV spectrometer due to the multiple internal components at high voltage (HV). The MCP within the Model 310G vacuum spectrometer is intended for operations at <1 × 10⁻⁵ Torr and, thus, risked arcing at the pressures found within the experiment (during a shot). The MCP was pulsed with HV to time-gate the experimental spectra. To protect the HV components, an extended beamline was designed with integrated dual-stage differential pumping (Fig. 2). Each differential pumping stage reduced the speed and pressure of any potential gas within the beamline by adding pumping volume and apertures (double-sided conflat blanks with circular holes machined into them) to constrict flow. The apertures increased in size moving away from the slit, toward the FuZE-Q plasma, and were 1.14 and 1.78 cm in diameter, respectively.

Between each aperture in the beamline were conflat tees with turbo-molecular vacuum pumps to evacuate each differential pumping section. In tests, a steady-state gas pressure of ∼10 mTorr was leaked into the mock beamline with dual-stage differential pumping leading to a reduction in pressure between the gas flow controller and the EUV spectrometer grating housing by a factor of ∼400.

A series of fast HV electronics were implemented in the EUV spectrometer diagnostic to further ensure the internal components would not be damaged by changes in pressure within the spectrometer housing. Importantly, all components of the detector including MCP strips and HV phosphor coated electrode were rapidly de-energized immediately after collecting data during each shot.

The MCP was time-gated with HV pulses to time-gate the experimental spectra. This was accomplished using Berkeley Nucleonics (model PVM-1001-N) HV pulsers that were triggered independently and used to time-gate the experimental spectra (pulse durations—gate widths—were between 55 ns and 10 µs), as shown in Fig. 3, bottom.

Furthermore, to decrease timing uncertainties associated with the collected spectra, the HV pulser output was digitized for later analysis. However, digitizing the pulser outputs along with the challenge of generating low-noise fast rise-time HV pulses on MCP strips required the pulsed HV circuitry to be carefully designed. In particular, the unterminated, 25 Ω impedance, MCP strips caused significant open-ended wave reflections in the coaxial cables during early circuit design testing. To limit the amplitude of the unwanted HV pulse reflections in the MCP and monitor circuits, a combination of high-pulsed-power resistors and attenuators was used. Each pulser was connected to a resistive network via a 50 Ω coaxial cable. R₁ represents a 50 Ω terminator [Fig. 4(a)], used in testing. During experiments, a 50 Ω, 20 dB high voltage pulse attenuator (ESDEMC model HVAT-3K20-3.5, with peak pulse-power rating of 180 kW) was used in series with a standard 40 dB attenuator, teed off the pulsed HV MCP circuit, allowing the HV pulse to be monitored. R₂ with 2 × 47 Ω (Vishay model DTO025C47R00JTE3, 25 W) in series [Fig. 4(a)] was added to mitigate the HV pulse reflection at the end of the MCP strip due to the impedance mismatch. This design produced satisfactory results with relatively clean pulse shapes and low noise, as shown in Fig. 3.

A novel feature of the implementation was the design of the HV circuits to pulse the phosphor coated thin-film electrode. The risk of arcing on the HV phosphor coated electrode was dramatically reduced using a “kill-switch” to safely short out the phosphor coated electrode and drop the voltage from ∼4250 to 0 V in ∼500 ns. The HV kill-switch used an N-channel HV MOSFET (IXTL2N470) rated to +4700 V with 40 ns turn-on time. Due to the MOSFET’s considerably high parasitic capacitance between the source and the gate, C_iss = 6860 pF, this capacitance had to be charged and discharged when turning the MOSFET on or off. This required a low impedance charge and drain circuit. This was accomplished by using a Microchip MIC44F18 MOSFET driver. The RC-time depended on the bleed resistor (1 kΩ) and the capacitance of the coaxial cable. The RC-time was measured to be ∼500 ns. See Figs. 4(b) and 4(c) for a schematic and image of the circuit.

Finally, to reduce the deposition of charged ion species onto the diffraction grating at the end of the beamline, a pair of permanent magnets was mounted on the backside of each aperture. These magnets impart a small force on all charged particles moving through the apertures, deflecting them slightly away from the entrance slit and grating, protecting the spectrometer.

As off-the-shelf steady-state EUV light sources were not easily available for calibration of this diagnostic, a custom microwave resonant cavity Helium (He) light source was built using an Opthos Evenson microwave cavity.²² The He light source coupled to the EUV spectrometer via a 1 mm diameter glass capillary, providing steady-state radiation between 50 and 60 nm. Using He I line emission as calibration data, along with device parameters such as the MCP offset and tilt, the pixel-to-wavelength conversion can be calculated for any MCP position along the Rowland circle.

Final wavelength calibration of the datasets included a correction calculated from the identification of oxygen line emission (see the  Appendix-Table I). The experimental spectra were compared with synthetic spectra from PrismSPECT collisional-radiative (CR) modeling²³ to confirm the line emission identification.

The values of Δy and γ calculated during final calibration are presented in Fig. 6. In addition, the He I emission spectra used for calibration in the 50–60 nm range is shown in Fig. 6(a). The He I (1s²–1s2p) emission line is particularly bright and anchors the 2p, 3p, and 4p series. The difference in wavelength between the peak (measured to the nearest CCD pixel) and the known database value is the wavelength error. If warranted, a sub-pixel peak-finding method will be implemented. The root mean square (rms) of the wavelength error [presented in Fig. 6(b)] is 0.011 nm, corresponding to ± one CCD pixel (∼0.01 nm).

The electron cloud exiting an MCP pore channel diffuses out radially, producing a decaying charge distribution and contributing to the observed quasi-exponential background. Most noticeable for very intense lines, such as the He I emission in Fig. 6(a), this effect is influenced by the electrostatic Coulomb repulsion of the electron cloud exiting the MCP pores.²⁵ As a result, the ratio of signal to background is noticeably lower (∼50:1) than expected when only considering the performance of the diffraction grating (a 1000:1 signal to background noise ratio is common for most diffraction gratings).

Several image corrections were implemented within the EUV data processing procedure. The image processing procedure reverses the mathematical transformations produced by the instrument components. The procedure is summarized first and then discussed in detail in the following paragraphs:

1. A CCD dark frame is subtracted from the raw image. This removes CCD artifacts, such as dark current, digitizer offset, and abnormal pixels.

2. MCP cross-pixel bleed is removed. An exponential spread function (f) [Fig. 7(a)] convolved with the dark-frame-corrected image (I₁) [Fig. 7(c)] is subtracted from I₁ [Fig. 7(b)]. The convolved background image is scaled by the intensity in the gap between the MCP strips.

3. Spectrometer-grating optical aberrations, such as coma, that distort the spectral line shapes slightly across the MCP are corrected [Fig. 7(b)].

The first step in the analysis is to subtract a dark frame from the raw image to eliminate CCD artifacts. Dark frame images have the same CCD exposure settings (mechanical shutter and CCD-chip collection period) as the raw image. This was typically 5 s but will be reduced in the future (the time-resolution of the experimental spectra is determined by the μs-pulsed HV MCP gate, not the CCD). Dark current levels were reduced dramatically from initial testing to final data collection by cooling of the CCD to 0 °C.

The second step is the removal of MCP cross-pixel bleed. A simplified visual summary of steps 2 and 3 is shown in Fig. 7. The vertical lineouts [Fig. 7(e)] illustrate the difference in the background across the image, before and after the correction.

The intensity and width of the spread function are best-fit parameters that may be refined further in future analysis. Previous studies found good agreement between the background intensity across the strips (perpendicular to the MCP strips) and a 1/(55 pixel) exponential spread function. The initial intensity of the spread function is arbitrary, but the convolved background image (the original image convolved with the spread function) is subsequently scaled. However, because the convolved background image, using a single, uniform scaling factor, does not closely match the spatially varying intensity in the dark gap between the strips, a horizontal-position-dependent scaling factor is calculated. The dark gap would have zero intensity were it not for cross-pixel bleeding, which is increased in bright locations by Coulomb repulsion (see Ref. 25). The convolved background image at each pixel location along the horizontal (wavelength) axis was scaled by this factor. Importantly, the scaling factor was determined not by the spectrum on the MCP strips, but rather by the dark background in the gap between the MCP strips. The two-step approach is:

1. The convolved background of the exponential spread function was scaled by the pixel values in the gap between MCP strips for each position along the wavelength axis of the image.

2. The scaled “slice” (for a column at constant wavelength axis position) of the convolved background was then subtracted from the corresponding slice of the image.

This processing technique improved the background correction of the EUV image data, especially at the edges of the image; see Figs. 7(b) and 7(d) for comparison. Finally, future analysis will include flat-field corrections to account for the MCP strip dependent sensitivity.

The second-generation FuZE-Q device is similar to FuZE, but with several important differences. First, the FuZE-Q pulsed-power supplies were significantly bolstered. Second, FuZE-Q introduced multiple changes to the gas-puff fuel delivery systems. Finally, changes were made to the materials used within the FuZE-Q device.

The EUV spectrometer was fielded on a 50°, 2–3/4 in. conflat vacuum port, looking at P20 (20 cm downstream from the cathode nosecone–close to the axial center of the FuZE-Q assembly region) (Fig. 8). The EUV spectrometer and beamline were electrically isolated from the FuZE-Q device with a ceramic break in the beamline. A metal bellows in the beamline protected the ceramic break from damage (torque from motion of the long beamline) and allowed for diagnostic alignment while under vacuum. The EUV spectrometer, along with the beamline apertures, was aligned to the FuZE-Q device (Fig. 9) with a low-power optical laser. Coarse alignment was completed by adjusting the position of the 8020 frame on which the EUV spectrometer was held in position with respect to the FuZE-Q device (Fig. 9). Fine alignment was achieved by adjustment of a two-axis stage with flexible bellows mounted to the EUV spectrometer immediately between the spectrometer and beamline.

With the alignment laser mounted on an adjustable stand opposite the EUV spectrometer on the other side of the FuZE-Q device, the laser beam was directed incident on the spectrometer slit and diffraction grating behind. A custom viewport with an optical mount was designed to hold a penta-prism (Thorlabs PS932-20 mm) inside the EUV spectrometer diffraction grating housing to allow side-on viewing of the grating and laser spot, aiding in alignment of the diagnostic.

The entrance slit width was commonly 25 µm in these experiments, with a height of 1.5 cm. The orientation of the slit with respect to the FuZE-Q plasma column was an important design consideration. In one scenario, a slit in the “horizontal” orientation, parallel to the plasma column, would likely see more of the plasma along the pinch axis. However, in another scenario, a “vertical” slit, as was chosen ultimately, would be more robust to fluctuations in the radial location of the plasma column, likely reducing shot-to-shot variability in the emission spectra intensity.

Emission spectra from the FuZE-Q device were collected at several positions along the Rowland circle with wavelengths between ∼5 and 40 nm. The experimental spectra were time-gated with HV pulses. The relative timing and duration of the HV pulses applied to the MCP strips determined the timing of the spectra and exposure duration. A range of timing and durations were used. The relative timings were generally set to study the emission before, during, and after the primary FuZE-Q pinch event (Fig. 10). The durations of the HV pulses applied to the MCP strips set the time-integrated exposure of a given spectrum. Considering that the timescale of plasma emission on the FuZE-Q device is >20 µs, durations of order 1 µs were generally chosen. However, spectra with exposures between 500 ns and 8 µs were also collected (Fig. 9).

The resolving power (λ/Δλ) of the spectrometer was calculated to be ∼600, in the ∼50–60 nm range, during calibration with the He UV light source. However, for the experimental data collected at shorter wavelengths, the resolving power decreased to ∼200 (Fig. 11). The lower resolution was partially due to the removal of an aperture mounted behind the slit that reduced the illuminated width of the grating, optimizing the spectrometer’s resolution. During initial experiments, the light throughput was of greater concern than the resolution, and therefore, this aperture was removed. The spectrometer was also operated with a larger entrance slit width than was used for initial calibration and resolution calculations.

Detailed analysis of the collected spectra is ongoing. As such, findings from the spectra shown are preliminary. The primary goal of the experimental campaign in late 2022 was to field the EUV spectrometer diagnostic and to demonstrate the diagnostic’s ability to study the time-dependent plasma impurity emission.

Early analysis identified oxygen as the dominant source of line emission in the FuZE-Q device. More specifically, N-, C-, B-, Be-, Li-, and He-like oxygen (O II–O VII) line emission was identified in many FuZE-Q pulses. The intensity of the oxygen line radiation decreased significantly (by a factor of $∼10$⁠) after the removal of a vacuum window in the line-of-sight of the EUV beamline. Oxygen line radiation is likely due to plasma–material interactions at the interface between the plasma and the vacuum window. These interactions release window material (predominately O and Si) into the vacuum vessel and plasma. However, oxygen line radiation returned to similarly high intensities in later experimental campaigns without the same vacuum window (replaced by a metal vacuum flange), suggesting that oxygen may also be sourced from plasma-window interactions, or other mechanisms, outside of the spectrometer’s line-of-sight.

Several observed emission lines do not appear to be associated with oxygen and are likely from carbon and silicon. Carbon and silicon are also identified in the UV spectra from another spectroscopy diagnostic.⁷ While line identification efforts are ongoing, and thus, the identifications presented are only suggested matches, the diagnostic is found to be capable of studying the time dependence of the plasma line emission on the microsecond timescale (Fig. 12).

The set of emission features shown in Fig. 12 is promising for future experiments and analysis that will examine the time evolution of specific impurities and emission features in FuZE-Q. Apart from these specific features, a broadband spectrum was collected over the wavelength range of 5–40 nm. Figure 13 presents emission spectra collected at six different locations along the Rowland circle, corresponding to the wavelength range ∼5–40 nm. From initial line identification efforts, it appears the spectrum is predominantly oxygen emission at a wide range of plasma temperatures. At this time, high-Z elements such as W, Cu, Cr, Zr, Mo, Ti, Ni, and Fe have not been definitively identified in the emission spectra.

The presence of oxygen, and other elements, was also confirmed with non-spectroscopic methods. O, Si, and C were found in deposition and debris within the FuZE and FuZE-Q devices. Scanning electron microscope (SEM) backscattered electron and energy dispersive x-ray spectroscopy (EDS) modes were used to analyze deposition and debris from within the FuZE vacuum chamber, and on a witness plate on FuZE-Q, collected after extended operation. These studies found O, Si, C, Cu, and W of various sizes and levels of homogeneity within the material in FuZE. On FuZE, it appears that tungsten embeds in a silicon–oxygen matrix in some larger debris particles. On FuZE-Q, O, Si, and C are the only elements identified in the SEM EDS analysis of witness plate deposition.

Oxygen line emission (N- through He-like oxygen) was observed across the broadband wavelength range. Detailed calculations of plasma temperature and density, including comparison to collisional-radiative synthetic spectra, are beyond the scope of this manuscript and are left for further publications. Future experimental work will use spectroscopic tracers such as Ne and Ar to permit more sensitive plasma temperature and density measurements. In addition, a 2400 g/mm diffraction grating may also be used, increasing resolution at shorter wavelengths.

A diagnostic for EUV spectroscopy was fielded on the FuZE-Q sheared-flow-stabilized Z pinch. Despite the challenges associated with direct line-of-sight, in-vacuum measurements, the spectrometer diagnostic collected high quality time-dependent emission spectra in the 5–40 nm wavelength (30–250 eV) range. Time-gated by pulsed HV systems, the diagnostic collected three-independently-timed emission spectra per pulse with 55 ns to 10 μs exposures. Early analysis identifies oxygen (N-, C-, B-, Be-, Li-, and He-like O) as the dominant element in the impurity emission. Future work will include analysis of plasma temperature, density, and other parameters such as Z_eff, supported by collisional-radiative modeling.

The authors would like to acknowledge and thank W. Cline, C. Davidson, A. Astanovitskiy, O. Dmitriev, K. Swanson, and R. Mancini from the University of Nevada, Reno. We also thank J. Coyne, W. McGehee, A. Hossack, A. Johansen, D. Austin, M. Parry, J. Smythe, L. Morton, E. Meier, P. Stoltz, K. Smith, and L. Pennings of Zap Energy, Inc. This work was supported by the U.S. Department of Energy through ARPA-E Award No. DE-AR0001161 and by Contract No. DOE-FOA-2212-1526 to Los Alamos National Laboratory (LANL). This work was also supported by Zap Energy, Inc. LANL is operated by Triad National Security, LLC, for the National Nuclear Security Administration of the U.S. Department of Energy (Contract No. 89233218CNA000001).

The authors have no conflicts to disclose.

Aidan Williams Klemmer: Conceptualization (equal), Data curation (equal), Formal analysis (equal), Investigation (equal), Methodology (equal), Visualization (equal), Writing – original draft (equal), Writing – review & editing (equal), Brian A. Nelson: Funding acquisition (equal), Project administration (equal), Supervision (equal), Tobin R. Weber: Investigation (equal), Supervision (equal), Morgan Quinley: Investigation (equal), Resources (equal), Uri Shumlak: Funding acquisition (equal), Project administration (equal), Supervision (equal), Writing – review & editing (equal), Stephan Fuelling: Conceptualization (equal), Formal analysis (equal), Investigation (equal), Methodology (equal), Resources (equal), Supervision (equal), Writing – review & editing (equal), Bruno S. Bauer: Conceptualization (equal), Funding acquisition (equal), Methodology (equal), Project administration (equal), Supervision (equal), Writing – review & editing (equal), Glen Wurden: Funding acquisition (equal), Project administration (equal), Resources (equal), Supervision (equal), Writing – review & editing (equal), Andrew S. Taylor: Formal analysis (equal), Investigation (equal), Writing – review & editing (equal), Derek A. Sutherland: Formal analysis (equal), Investigation (equal), Writing – review & editing (equal), Akash P. Shah: Formal analysis (equal), Investigation (equal), Writing – review & editing (equal), Anton D. Stepanov: Formal analysis (equal), Visualization (equal), Writing – review & editing (equal), Benjamin Levitt: Funding acquisition (equal), Project administration (equal), Supervision (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Table of known oxygen and helium emission lines used for calibration at each MCP position along the Rowland circle (5.7–11 in.) (Table I).